Journal of Pragmatics 36 (2004) 1–24

www.elsevier.com/locate/pragma

Introducing polylogue
Catherine Kerbrat-Orecchioni
Groupe de Recherches sur les Interactions Communicatives§,
CNRS-Université Lumière Lyon 2, 5 av. Pierre Mendès France,
69676 Bron, France

Received 26 February 2002

Abstract
The introduction to this special issue begins by defining the notion of ‘polylogue’. Then,
after having summarized the results of our previous work on ‘trilogues’, I propose a survey of
the general perspective adopted by the authors, and of the main analytical tools they use.
Finally, the articles gathered in the volume are introduced in more detail in relation to the
particular situations and data they deal with.
# 2003 Elsevier B.V. All rights reserved.
Keywords: Dilogue/trilogue/polylogue; Plurilevel analysis; Typology of polylogues; Participation framework

1. Criticism of dyadic communication models

Dyadic communication is widely thought to be the communicative situation par

excellence—not only by linguists, semioticians, psychologists, and communication
theoreticians, but also by ‘the man on the street’, as witnessed, for example, by the
fact that the word dialogue, despite its etymological origins,1 is generally understood
to mean ‘conversation between two people’. This can, of course, be explained by the
confusion between the two paronymous prefixes di- and dia-, but is doubtless also

§
All the authors of the articles composing this issue belong or are associated to this research team,
working in Lyon (France). The different texts which are collected here must in fact be considered the
result of a collective research project. Some of the articles were originally written in French, some others
directly in English. All the data we analyse was originally produced in French. The whole text was trans-
lated or edited by Louise Nicollet, whose thoroughness we are sincerely grateful for. Many thanks also to
Dick Janney for his encouragement, his patience, and his perfectionism in the revision process.
E-mail address: kerbrat@univ-lyon2.fr (C. Kerbrat-Orecchioni).
1
Since the Greek prefix ‘dia-’ means not ‘two’, but ‘through’. In order to avoid ambiguity, we prefer to
speak of ‘dilogue’ when referring to exchanges between two people.

0378-2166/03/$ - see front matter # 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0378-2166(03)00034-1
2 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

due to the deep-rooted tendency to associate interaction with interaction between

two people, considered as the prototype of all forms of interaction.
For some time now, this ‘privilege’ in favour of dyadic communication has been
severely criticized by numerous researchers such as Hymes, Goffman, Levinson, and
specialists in the field of Conversation Analysis (henceforth, CA), e.g.:

The common dyadic model of speaker-hearer specifies sometimes too many,

sometimes too few, sometimes the wrong participants. (Hymes, 1974: 54)

Traditional analysis of saying and what gets said seems tacitly committed to the
following paradigm: two and only two individuals are engaged together in it,
[. . .] the two-person arrangement being the one that informs the underlying
imagery we have about face-to-face interaction. (Goffman, 1981: 129)

In the study of verbal interaction, there has been undoubtedly some bias
towards the study of dyadic interaction. (Levinson, 1988: 222–223)

Levinson even speaks in this connection of a ‘straightjacket’, and he shows that

‘dyadic triumph’ has been achieved at the price of greatly limiting, first, the situa-
tions which are examined (in any society, dyadic exchanges tend, in fact, to be in the
minority), second, the cultures under consideration (many societies accord an even
more important role than Occidental societies do to ‘multi-party gatherings’2 and to
all sorts of relayed or ‘mediatized’ communication). So Levinson regards this dyadic
diktat as ethnocentristic.
As for CA, in their seminal article on the turn system, Sacks et al. (1974/1978)
claim that their ‘simplest systematics’ of turn-taking is applicable to all conversa-
tions, no matter how many participants are involved. But at the same time, they
recognize that ‘‘numbers are significant for talk-in-interaction’’, as the title of a more
recent article by Schegloff (1995) recalls. The conversational data used by CA spe-
cialists is in fact diversified in this respect.3 However, it cannot be said that analyses
in this field have exhaustively covered the topic in which we are interested, that is,
the description of all the phenomena which characterize the functioning of poly-
logues. First, the situations looked at by CA correspond to focused interactions
whose participation formats and consequent functioning are relatively simple (in
any case much simpler than those of the interactions which will be examined here).
In addition, CA concerns itself mainly with local phenomena such as the turn-system,

2
Also see Aronsson (1996), who asserts that in many non-Occidental cultures, basic communication
situations are of ‘polylogal’ type (e.g. in traditional African societies, the mother–child dialogue generally
takes place in the presence of siblings or other members of the family).
3
In his Lectures on Conversation, Sacks already bases his observations mainly on a set of data com-
posed of a therapy session bringing together an adult therapist and a group of teenagers. Concerning
SSJ’s article on turn system, O’Connell et al. (1990) note, however, that 71% of the 35 examples men-
tioned only involve two speakers; and that the large majority of empirical studies carried out in this per-
spective (exactly 82% out of a corpus of 22 publications—the sample is therefore limited, and stops in
1990) is based on ‘dilogal’ data.
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 3

whereas many other aspects can be taken into account that are affected even more by
the number of participants. Finally, and above all, for proponents of CA, turn-taking
operates not between speakers but between ‘parties’, with speakers considered only as
being ‘incumbents’ of these parties. Schegloff (1995: 32–33),4 for example, claims that
‘‘the turn-taking system as described in SSJ organizes the distribution of talk not in
the first instance among persons, but among parties.’’ This can involve:

their relative alignment in current activities, such as the co-telling of a story or

siding together in a disagreement, or their several attributes relative to a
momentarily current interactional contingency, for example, whether they are
host or guest, whether—as a new increment is being added to a number of
interactional participants—they are the newly arrived or pre-present.

Thanks to this notion of party, it is possible ‘‘to introduce order into this potentially
chaotic circumstance’’ constituted by the large number of participants. (1995: 40)
For us, on the contrary, turn-taking operates per se between speakers. Even if
mechanisms of alignment based on statuses, roles, speakers’ objectives, etc. play an
important role in conversation, the succession of turns is first and foremost a phe-
nomenon which takes place between individuals. The notion of ‘party’ (which, in point
of fact, covers diverse phenomena) belongs to another level of analysis, as we will see.

2. The notion of polylogue

For the reasons discussed above, in designating the topic dealt with in this pub-
lication we will not speak of ‘multi-party conversations’ but of multi-participant
conversations, or rather multi-participant interactions (conversation, in the ordinary
sense, being only one particular type of talk-in-interaction). We will also speak of
polylogues.5 This term is etymologically appropriate, it fits into a coherent paradigm
(‘dilogue’, ‘trilogue’, ‘tetralogue’ etc.), and it is easy to handle, allowing, for example,
the derived adjective. Thus, we will refer to as polylogal all communicative situations
which gather together several participants, that is, real live individuals. Thus defined,
the notion of polylogue, although seemingly trivial, already poses some problems
owing to the difficulty of clearly defining the category of ‘participants’. I will come
back to this point, but for the moment, let me say that the situations analysed in this
issue involve variable numbers of participants, ranging from four (e.g. the interac-
tions involving a divorced couple and their respective notaries studied by Bruxelles

4
And before him, Sacks (Lectures vol. I, 523–524): ‘‘two parties does not necessarily mean two persons’’.
5
In a completely different sense from the meaning attributed to this term by Julia Kristeva (in Poly-
logue, Paris: Seuil, 1977). Picking up on Maurice Blanchot’s idea of ‘parole plurielle’, Kristeva’s approach
is related to what is commonly known as ‘polyphony’, or ‘dialogism’. To use the terminological distinc-
tion (introduced by Eddy Roulet) between dialogal discourse (which brings together several distinct
speakers) and dialogic discourse (which refers to a plurality of enunciative voices, more abstract entities
which can be embodied by one and the same speaker), we will say that Kristeva’s perspective is ‘dialogic’,
whereas our perspective in this issue is rather ‘dialogal’.
4 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

and Kerbrat-Orecchioni) through various intermediate situations (e.g. the hospital

shift-change briefing sessions studied by Grosjean) all the way to a theoretically
infinite number of participants (e.g. the Internet newsgroups studied by Marcoccia).
‘Trilogues’, the minimal form taken by polylogues, were eliminated from this study
because our research team has previously carried out work on these configurations;
here, I will just briefly outline the objective and the results of that earlier work.6

3. Trilogues

The objective of our earlier studies was to bring to light the specific features of
trilogues that distinguish them from dilogues. Trilogues were studied at every level
of their functioning, through a large range of data (everyday conversations,
exchanges in the media, and talk in different institutional contexts).

3.1. The hearers’ roles

We were particularly interested in studying the distinction between an ‘addressed

recipient’ and a ‘non-addressed recipient’, and the related question of addressing
cues, which are most often fuzzy. Given the continuity between these two categories,
it is preferable to talk of main addressee vs. secondary addressee. In our studies, we
constantly observe fluctuations in address. Goodwin (1981) shows very clearly that,
along with the change in addressee, the pragmatic value of an utterance can be
modified in the course of the exchange (an initially informative segment being, for
example, reconverted into a demand for confirmation, or vice versa).
An utterance can also simultaneously convey different pragmatic values for its differ-
ent hearers. This idea was already put forward by Sacks (Lectures, vol. I: 530–534 and
vol. II: 99–101), who stated that an utterance addressed to B can very well ‘‘do some-
thing’’ to C that is different from what it does to B (if, for example, A flirts with B, ‘‘then
she may be teasing C’’). A similar idea is elaborated by Clark and Carlson (1982: 333),
who note that ‘‘speakers perform illocutionary acts not only toward addressees, but
also toward certain other hearers’’ (an order, for example, can have the value of an
informative act for ‘lateral’ hearers, and possibly even some additional values).

3.2. The organization of turn-taking

Turn-taking7 in trilogues is generally characterized by variability in alternation

patterns. This variability is in part a result of lack of balance in floor-holding,
violations of speaker-selection rules, and interruptions and simultaneous talk.
6
Results which led to a publication (Kerbrat-Orecchioni and Plantin, 1995); for a synthesis of this
research project, see Kerbrat-Orecchioni (1997).
7
See Sacks et al. (1978: 23), for the particular case of three-party conversations. For more or less
radical (and more or less justified) critiques of this approach, particularly applied to complex participation
formats, see Edelsky (1981), Raffler-Engel (1983), Power and Dal Martello (1986), O’Connell et al. (1990),
Ford et al. (1996).
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 5

3.2.1. General lack of balance in floor-holding

In dilogues, only unequal lengths of turns make for disproportionate amounts of
participation in the interaction. In trilogues, on the other hand, unequal participa-
tion is first based on the number of turns, and it is always possible for a participant
to be left out, or to temporarily eliminate him- or herself from the exchange, for
different reasons and with varying effects. This possibility increases where the parti-
cipants are more numerous.8

3.2.2. Violations of speaker-selection rules

In the selection of the next speaker, trilogues often feature violations of con-
versational rules that are unknown in dilogues. Intrusions (which happen when the
participant who takes the floor is not the one who has been selected by the current
speaker) are frequent, for example, in interviews of couples (Marcoccia, 1995) and in
situations involving professionals (doctors, judges) and children accompanied by
their parents. It should be noted that such intrusions do not always constitute real
conversational offenses. According to Dausenschön-Gay and Krafft (1991: 148–149),
they sometimes follow an ‘efficiency rule’ which overrides Sacks et al. selection rule:

If for example I ask A what time it is, I will be completely satisfied if it is B who
answers me. Or to stay with the example of ‘Le masque et la plume’ [‘The Mask
and the Quill’]:9 there are times when Bastide [the moderator] mistakenly asks for
a commentary or explanation from one of the critics who has nothing whatso-
ever to say on that subject. As is quite normal, another critic answers and no one
even entertains the thought of taking offense.’’ (translated from the French)

3.2.3. Interruptions and simultaneous talk

The frequency of interruptions and simultaneous talk, as well as the variety of
ways in which these are carried out, increases in trilogues, and a fortiori in multi-
participant interactions. In Müller’s (1995) study of discussions among eight French
students, for example, there are constant overlaps of three or even four voices
superimposed over each other. The first impression created by the overlaps is one of
unbearable cacophony, but deeper analysis reveals the concerted organization of
these interruptions, which more often than not have a collaborative function.
To summarize, multi-participant conversations are both more conflictual (there
are more opportunities for a struggle for the floor, and for violations or failures in
the functioning of the turn-system) and more open to mediation and conciliation

8
The potentially destabilizing presence of a silent participant is superbly illustrated by Nathalie Sar-
raute’s play Le Silence (1967), based on the following situation (Paris: Folio, 1998, Abstract p. 93; trans-
lated from the French): ‘‘Six people—or rather six voices—find themselves unable to pursue a ‘normal’
dialogue due to the silence of a seventh person. [. . .] Why does Jean-Pierre remain so obstinately silent?
Why doesn’t he answer when someone asks him a question? What is he thinking? Does he pass judgment
on his more talkative partners? Is he hostile? indifferent?’’—all these puzzling and unanswered questions
will end up, as the play shows, undermining the conversation.
9
A discussion on a French radio program, analysed by these authors.
6 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

than dilogues. They also place fewer constraints on participants, since the obligation
to cooperate—being in a way ‘diluted’ by the larger group—is not as strong for each
individual speaker.

3.3. The structuring of dialogue

The structuring of interaction into hierarchically organized units (monologal

‘moves’, dialogal ‘exchanges’, etc.) in trilogues is also different from that of dilogues:

 Within one and the same exchange, the initiating move, like the reacting
move, can be composed of several contributions produced by different
speakers.
 The question of the completeness/incompleteness of exchanges is posed in
different terms, and the phenomenon of ‘truncation’ takes different forms.
 The structural organization of trilogues is clearly more complex than that of
dilogues (intertwining of exchanges, conflicts over structuring, etc.).

The complexity is obviously even greater in configurations with more participants.

Delamotte-Legrand (1995), for example, studied the division of exchanges in dis-
cussions within groups of nine pre-adolescent children, and found that exchanges
(defined as a succession of moves dependent on a single initial move) with 3–5
speakers and 4–8 moves were dominant. These represented more than half the
exchanges in the data studied. Delamotte-Legrand’s findings suggest that, when we
are dealing with configurations as complex as these, it is necessary to re-think the
‘canonic’ structure of the exchange as well as the notion of ‘truncation’.

3.4. The level of interpersonal relationships

With regard in particular to the relation of dominance among the participants, the
trilogue allows certain members of the triad to form coalitions, a notion developed
by Theodor Caplow which will be investigated later (see Bruxelles and Kerbrat-
Orecchioni, this issue).

In conclusion, for the analyst, the functioning of trilogues is in all regards more
complicated to describe than that of dilogues. For the participants themselves, the
more numerous they are, the more delicate conversational activities become.
Speakers must take all their recipients into account to some degree, and the recipients
themselves are intrinsically heterogeneous due to differences in status, knowledge,
expectations, objectives, etc. This situation can lead to apparently contradictory
utterances, as Müller (1997: 386) points out in his comment on Lonardi and Viaro’s
(1990) work on interviews between therapists and their patients:

In contradictory situations of this kind, members respond by contradicting

themselves, responding in a confused manner, pretending to have forgotten. Ana-
lyzing in detail such a contradictory response, LV suggest that these contradictions
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 7

can be read as cases of ‘multiple recipient design’, i.e. the speaker attempts to
tell, within one and the same turn, a proposition A to the therapist, but non-A
to his co-present wife.

When a triad is conversing, the moments of ‘genuine’ trilogue (in which all
three members are actively engaged) never last long; instead, they alternate with
phases which seem rather dilogal in character, involving two active speakers and
a third participant who can adopt various attitudes and show extremely variable
involvement in the interaction in progress. Describing the interaction therefore
requires above all observing gradual shifting from dilogal structures to trilogal
ones and vice versa—observing, that is, the mechanisms of ‘connecting’ and ‘dis-
connecting’ the third participant. In any case, these moments of ‘dilogues within
trilogues’ must not be dealt with as real dilogue, since they take place in the
presence of a third party; this fact is always relevant and should be taken into
account.10
The main characteristics of trilogues, in comparison with dilogues, are their
flexibility, instability, and unpredictability, which are identifiable at all levels of
their functioning: this is the major conclusion to which our observation of trilogues
has led. Flexibility is, of course, even greater in interactions involving more
numerous participants. Beginning with four participants, a new possibility appears:
‘splitting-off’, that is, the forming of distinct conversational groups which continue
parallel exchanges (see Traverso, this issue). Beyond four participants, the pro-
blems of describing the interaction increase dramatically (especially in case of
informal non-focusing interactions). The first impression created by an audio
recording of such an interaction is one of such confusion and anarchy as to dis-
courage any attempt at analysis. . . But then, via immersion in the data, islands of
organization and regularity begin to emerge, as we will try to show in the following
analyses.

4. Principles of analysis

Our study of polylogues is based on the same principles which we developed for
describing trilogues.

4.1. Levels of analysis

We distinguish among several levels of analysis, and several types of units. For
example, we consider ‘turns’ and ‘moves’ to be two different kinds of monologal
units.

10
For example, as pointed out by Brown and Levinson (1987: 12), the seriousness of a FTA (Face
Threatening Act) increases in the presence of third parties; and Paddy Austin goes so far as to state (1987:
20): ‘‘An individual’s face is vulnerable in direct proportion to the number of people to whom she presents
that face in any given interaction.’’
8 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

*Turns (as well as Turn-Constructional Units) are ‘‘epiphenomena’’ (Selting, 2000:

511), that is, ‘‘surface units’’, cf. Roulet (1992: 92–93):

In fact, a turn is a unity which pertains to the surface structure of conversation,

since it is uniquely marked by a change of speaker and does not necessarily coincide
with the speech activities (like questions, answers, requests, etc.) of the speakers.

A turn is produced by only one speaker, but it can have several successive recipients
and hence it can be divided up into different utterance-events (Levinson, 1988).11
These segments are sometimes very short, as shown by Goodwin’s (1981) example ‘‘I
gave up smoking cigarettes one week ago today, actually’’, which is composed of three
successive ‘utterance-events’ (sections according to Goodwin’s terminology).

*Moves, on the other hand, correspond to activities that participants intend to

accomplish via turns. They pertain to another level of analysis, constituted by
pragmatic units that belong to different ‘ranks’:12

4.1.1. Upper rank: conversation, or, more generally, interaction

Conversation is traditionally assumed to be, as in Goffman’s definition (1981: 130):

a substantive, naturally bounded stretch of interaction comprising all that rele-

vantly goes on from the moment two (or more) individuals open such dealings
between themselves and continuing until they finally close this activity out. The
opening will typically be marked by the participants turning from their several
disjointed orientations, moving together and bodily addressing one another; the
closing by their departing in some physical way from the prior immediacy of
copresence. Typically, ritual brackets will also be found, such as greetings and
farewells, these establishing and terminating open, official, joint engagement,
that is, ratified participation. In summary, a ‘social encounter’.

But Goffman adds that this definition is in some cases totally contradicted, par-
ticularly in discontinuous situations characterized by an ‘open state of talk’, or in
multi-focus settings where different encounters are closely intertwined with each
other. Although in many cases the largest unit is clear-cut, in others, it is not, and
the investigator is wise to drop the idea of even attempting to describe a global
interaction, and instead be content with taking a more modest approach and exam-
ining only certain ‘moments of talk’. At any rate, it is at this ‘macro’ level that the
notion of script comes into play. Schank and Abelson define a script as ‘‘a structure
that describes appropriate sequences of events in a particular context’’, that is ‘‘a
predetermined, stereotyped sequence of actions that defines a well-known situation’’

11
An ‘‘utterance-event’’ is for Levinson (1988: 168) ‘‘that stretch of a turn at talk over which there is a
constant set of participant roles mapped into the same set of individuals’’.
12
The principle of rank-analysis was developed by various discourse analysts, such as Sinclair and
Coulthard (1975), Edmondson (1981), or Roulet (1981).
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 9

(1977: 42–43). They add that ‘‘a script must be written from a particular role’s
point of view’’; for example, what we commonly call the ‘Restaurant Script’ is, in
fact, a script of waiter–customer interaction as seen from the customer’s point of
view.

4.1.2. Intermediate rank: episodes, phases, sequences

These terms designate blocks of exchanges which possess a high degree of thematic
or pragmatic cohesiveness, and whose configurations vary according to the type of
interaction.

4.1.3. Lower ranks: exchanges and moves

Exchanges are defined as the smallest pragmatic units to be produced by at least
two different speakers. Moves correspond to the contribution a given speaker makes
to a given exchange.13

4.2. Descriptive tools

Since the same tools are not appropriate to describing the different levels and
components of interaction, it is necessary to call upon various descriptive traditions.
Thus, our analyses will occasionally make reference to Searle’s or Grice’s prag-
matics, Hymes’s ethnography of speaking, Gumperz’s sociolinguistics, social psy-
chology, cognitive approaches, etc. But our basic kit of tools is supplied more
specifically by:

 Goffman’s theory of the participation framework, and his description of the

interpersonal relationship level, in particular, face-work phenomena;14
 Discourse Analysis (‘Birmingham school’ or ‘Geneva school’) for an inter-
action’s internal organization, seen in relation to its different ranks; and
 Conversation Analysis, for all aspects regarding local arrangements.

We believe that there is no incompatibility among these different theoretical

approaches, as they do not compete on the same turf.15

13
In this perspective, exchanges (such as the adjacency pair ‘question-answer’) are in fact combinations
of moves and not of turns. A turn can be composed of several moves, and therefore be part of several
exchanges. Inversely, one and the same move can (more exceptionally) be distributed over several turns,
for example in case of co-énonciation (Jeanneret, 1991, 1995, 1999).
14
These phenomena are described more precisely by Brown and Levinson (1978, 1987), whose theory
of politeness is of great use to interaction analysis, for example in order to describe the mechanisms of
preference organization (Lerner, 1996).
15
This ‘eclecticism’ characterizes most of the French research work in analysis of interactions, but it is
also claimed by some Anglo-Americain researchers, like Aston (1998: 13): ‘‘Since our primary objective
was not that of testing a specific discourse theory, however, our approach to the description of the data
has been substantially eclectic; the different theoretical backgrounds of the various members of the group,
and the nature of our aims, have entailed that rather than opting for a single descriptive model a priori, a
series of models have been examined, with a consensus as to requirements gradually emerging.’’
10 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

5. Problems of typology

Polylogues can be of very diverse nature. Among the classification criteria on which
a typology can be established, there is first of all the number of participants, which
can be extremely variable. However, as Grosjean and Traverso state (1998: 51):

The question of number is not in itself fundamentally the main question.

Although they bring together the same number of people, the following situa-
tions have hardly any aspects in common: eight people in a waiting room, eight
friends sharing a meal, eight individuals who work in the same office. (trans-
lated from the French)

In this same article, Grosjean and Traverso bring other relevant axes to light, one
of these being the focused or unfocused nature of the interaction being observed. On
this topic, they introduce a number of useful distinctions which sharpen that of
Goffman between simple ‘gatherings’ and veritable ‘encounters’, where the group
forms around a common focus of attention.

5.1. Shared focus encounters

In shared focus encounters, the different participants are oriented towards one and
the same activity, verbal or non-verbal. These interactions can take place in a formal
or informal frame, formality being a gradual phenomenon (Drew and Heritage,
1992a: 27; see also Grosjean’s and Traverso’s analyses of ‘semi-formal’ situations, this
issue). The less formal the situation is, the more phenomena of ‘splitting-off’ can be
observed. Their highest level of frequency is to be found in everyday conversations
among friends.

5..2. Unfocused gatherings

The classic example of an unfocused gathering is the waiting room, where non-
involvement is the rule, but copresence in an enclosed space brings about a situation
of ‘latent communication’.16

5.3. Multi-focus gatherings

Multi-focus gatherings are situations in offices, workshops, etc. where the different
participants or groups of participants go about different activities in the same place:

Under these conditions, an ‘open state of talk’ can develop, participants having
the right but not the obligation to initiate a little flurry of talk, then relapse

16
A situation which is managed in very diverse ways depending on the culture: unlike the way things
work in France, in some societies, copresence in the same place almost automatically leads to beginning
conversation.
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 11

back into silence, all this with no apparent ritual marking, as though adding but
another interchange to a chronic conversation in progress. (Goffman, 1981:
134–135)

In professional settings, multi-focusing is ‘structural’. In more informal situa-

tions, where participants are pulled to and fro between different foci of attention, it
can be said to be ‘emergent’, as in the double-focus and even triple-focus interac-
tions in the home described by Vincent (1995) and Grosjean and Traverso (1998:
62–63). For each of the situations earlier, there is a corresponding participation
framework.

6. Erving Goffman’s notion of participation framework

Goffman (1981: Chapter 3) contributes the three following important notions:

*Production format: the speaker can take on the roles of ‘animator’, ‘author’, or
‘principal’.17 In addition to this, he or she can adopt variable footing; for example,
Clayman (1992) shows that in news-interview discourse, the interviewer shifts
from ‘neutral’ footing (consisting of producing purely factual utterances) to an
‘evaluative’ or even ‘controversial’ attitude.
*Participation framework:

When a word is spoken, all those who happen to be in perceptual range of the
event will have some of participation status relative to it. The codification of these
various positions and the normative specification of appropriate conduct within
each provide an essential background for interaction analysis. (Goffman, 1981: 3)

The participation framework corresponds to the ensemble of ‘participation statuses’.

*Participation status: Goffman distinguishes between ‘ratified participants’
(‘addressed’ and ‘unaddressed recipients’) and ‘non-ratified participants’
(‘bystanders’: ‘overhearers’ and ‘eavesdroppers’). Curiously enough, he does not
refer in this connection to a ‘reception format’. In our approach, in order to
restore symmetry, we differentiate within the participation framework between the
production format and the reception format.

These very useful distinctions nevertheless bring up a number of difficulties.

6.1. Problems

6.1.1. Participation
Participants are defined by Goffman as being ‘‘in perceptual range of the event’’.
But the criterion of perceptual access (visual and/or auditory) is too limited, because

17
See Marcoccia (this issue).
12 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

it excludes situations such as written discourse, relayed talk, etc. It is also too broad,
in the viewpoint of some researchers. In fact, for Goffman, every person present on
the site of the interaction, whether officially ratified or non-ratified as a participant,
has ipso facto a certain ‘participation status’, insofar as this copresence cannot fail to
have some impact on the behaviour of the people who are in contact with each other.
Levinson (1988), however, in opposition to Goffman, defines only those who are
‘ratified’ and ‘attentive’ as ‘participants’. And Goodwin (1981: 107 ff), noting a con-
tinuum between evident engagement and total disengagement (and vice versa), adopts
a kind of intermediate position, suggesting that even those who are ‘momentarily
disengaged’ should be included in the category of participants. It seems therefore
advisable to recognize the existence of different degrees of participation, according to
the scheme proposed by Bell (1984):

addressee: known, ratified, addressed (maximal degree);

auditor:18 known, ratified, unaddressed;
overhearer: known, unratified, unaddressed; and
eavesdropper: unknown, unratified, unaddressed (minimal degree).

6.1.2. Ratification
Ratified participants are, according to Goffman, officially a part of the conversa-
tion group, as witnessed by the way the members of the group are physically posi-
tioned (proxemics, postures, eye-contact network). Yet there is a great deal of
disagreement among theoreticians regarding this issue as well. For Drew (1992), for
example, jurors in courtrooms are considered ‘overhearers’ since they are forbidden
to speak; Heritage (1985) has the same attitude regarding the ‘audiences’ of news
interviews; and McCawley (1984: 263) contributes a very slight nuance to this view,
claiming that ‘‘the jury are only very loosely speaking ratified recipients [. . .]. The
spectators are not, even in a loose sense, ratified recipients, though they in many
cases are intended recipients.’’ Nevertheless, if seems difficult not to admit that jur-
ors and even spectators of a trial in a courtroom are ratified to a certain extent,
given the legitimacy of their presence on the site, their displayed interest in the pro-
ceedings, and the fact that the discourse produced is also (sometimes even mainly)
intended for them. Indeed, it seems preferable to recognize that there are different:

*degrees of ratification: We can agree with McCawley that the jurors are ‘more
ratified’ than the spectators, who are nevertheless still ratified;
*modes of ratification: Complete ratification (in both the production and reception
formats) can be opposed to ratification as a listener only (ratification in the
reception format). The latter is the case of all kinds of ‘audience’, such as the
spectators attending a court trial, a town council meeting, or a TV talk show,19

18
Indeed, this is the way Bell designates Goffman’s category of ratified non-addressed recipients.
19
In the last case, there are two categories of spectators, those who are present in the television studio
and those who are watching the program on their TV screens. In a sense, both are ratified as listeners.
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 13

whose permissible reactions are limited to a few non-verbal manifestations such as

laughter or applause;
*levels of ratification: On a global level, ratification is based on the script, which
defines official roles; however, some types of ratification (and non-ratification) are
only determined at a local level, in relation to a particular episode or task (see
Bruxelles and Kerbrat-Orecchioni’s analysis (this issue) of a radio discussion in
which, on a global level, all the participants are ratified, but, depending on the
sequence, one participant or another will be officially ratified by the moderator);
and
*ratification devices: Various devices can be used by the ratifier, or by the
participant seeking ratification (indicators, mainly kinesic, of attentiveness
and involvement). These techniques vary depending on the interaction
situations.20 Ratification may of course give rise to negotiating among the
participants.

6.1.3. Address
Within the category of ratified participants, Goffman distinguishes between
addressed recipients (to whom the speaker officially addresses his or her utterance)
and non-addressed recipients (‘side participants’ for Clark, ‘auditors’ for Bell). The
determination of addressee(s), however, poses the problem of address markers. As
explicit signals (principally terms of address) are rather seldom present, we most
often have to deal with subtle and gradual cues such as the content of the utterance
(which more specifically ‘concerns’ a particular listener), or paralinguistic and kine-
sic indications (vocal intensity, intonations, eye and body orientation, head move-
ments). Goffman insists in particular on ‘visual cues’,21 and defines the addressee as
‘‘the one to whom the speaker addresses his visual attention’’ (1981: 133). But such
markers are often ambiguous, and they may even be contradictory with each other
(examples of a clash between verbal and non-verbal markers: a sweeping glance
accompanying an utterance in the second person singular or just the opposite).22 So,
instead of referring to a discrete opposition between ‘addressees’ and ‘non-addres-
sees’, it seems preferable to assume that address cues often establish a gradual
ranking of main addressee(s) and secondary addressee(s). This continuum can have
two forms.

6.1.3.1. Continuum at a moment t1. At any given t1 moment of the interaction, a

recipient can have a status that is intermediate, i.e. between that of an addressed and
a non-addressed recipient. Cases of this would be:

20
In the example of the town council meeting (Witko, 2000), the roll-call procedure constitutes the
main ratification technique.
21
Inaccessible to the analyst who does not have access to a video recording. . . We note here that the
problem is posed in completely different terms when the data being studied is written: see in this issue
Marcoccia’s study on a case of communication via the Internet.
22
For an example during a corporate work session, see Lacoste (1989: 266–267).
14 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

*Semi-self talk: This is the case when it is not certain whether the speaker is talking to
him- or herself or to someone else.23 Well-known examples are interjections, excla-
mations, ‘response cries’ (cf. Goffman, 1981: Chapter 2), and ‘out-louds’ (cf. Levinson,
1988: 206 ff). There are also cases in which someone thinks aloud in someone else’s
presence, as in private comments by a customer in a cafeteria line, or by an office
employee working in front of a computer; and in domestic situations where each
participant goes about his own activities, producing brief apparent soliloquies, which
are not in fact authentic ‘self talk’, since the presence of other people exercises a certain
degree of control over the vocal productions of all the participants (see Vincent, 1995).
*Collective address: This is the case when a speaker is talking before a large
audience, as in a classroom setting, for example, where the teacher’s sweeping
glance over the audience is perforce unequal, ‘favouring’ some members (because
they perhaps produce more back-channel signals), although still not exactly
leaving the other members of the audience ‘unaddressed’.

6.1.3.2. Continuum between t1 and t2. Between t1 and t2 (two consecutive moments
in time), a gradual shift can occur from one addressee to another. For example, the
main addressee can be changed in mid-sentence by shifts of gaze, subtle vocal var-
iations, etc. This can even be done more brusquely by using the pronominal marker,
as in the following two examples (one of which is excerpted from a television debate
and the other from an informal conversation):

(1) He doesn’t want to be reduced to one of the two aspects of your personality.
[co-reference between ‘‘he’’ and ‘‘your’’, in relation to a change in the orientation
of the speaker’s glance]
(2) Since you stopped teaching the class, she doesn’t know how to talk!

6.1.4. Non-ratified recipients (or bystanders)

Non-ratified recipients, or ‘adventitious participants’, as Goffman notes, are pre-
sent in most communication situations (‘‘Their presence should be considered the
rule’’). Unlike ratified recipients, they must theoretically (pretend to) be disinterested in
what is going on within the conversation group. However, this is not always what really
happens. Goffman distinguishes between two sub-classes within the category of
bystanders: overhearers and eavesdroppers. This distinction is, in fact, based on two
different criteria which do not necessarily converge: the speaker’s awareness/unaware-
ness of the bystander, and the intentionality/unintentionality of the bystander’s hearing.
The criterion most often applied to distinguishing between overhearers and
eavesdroppers is whether the speaker is aware or unaware of the presence of the

23
‘‘She said something in a muffled voice, in a murmur, but it was difficult to know whether she was
saying that for his benefit or to herself.’’ (Milan Kundera, Risibles amours, Paris: Gallimard, 1974: 180;
translated from the French). The ‘semi-self talk’, or ‘half-aside’ (semi-aparté) is a frequent theatrical
device, particularly in Molière’s comedies.
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 15

bystander within the perceptual space. On the basis of this criterion, examples of
overhearers would be: factory workers doing repair work in an office where a meeting
is being held, or staff in charge of handling technical problems during a conference.
Such participants are only exceptionally promoted to the status of ‘addressees’.
Examples of eavesdropping, on the other hand, would be hearing a private conversa-
tion through a half-open door—or listening to a recording of a conversation which
just happens to have fallen into your hands as a conversation analyst. . .
In addition to this speaker-linked criterion, Goffman’s definition brings in another
criterion, connected to the attitudes or motives of hearers themselves. Overhearers,
according to Goffman, follow the talk temporarily, unintentionally, and inadvertently,
enacting shows of disinterest and minimizing their actual access to the talk. Eaves-
droppers, on the other hand, are indiscreet listeners who do everything they can to
intercept discourse which is in no way intended for their ears; in Goffman’s words,
‘‘they may surreptitiously exploit the accessibility they find they have.’’ (1981: 132).

6.2. Proposals

6.2.1. Main distinctions

Within the participation framework (ensemble of participation statuses), we will
base our study of polylogue on the following distinctions:

 Production format vs. reception format;

 Ratified participants vs. non-ratified participants (with, as we have seen, dif-
ferent degrees and modes of ratification);
 Main addressees vs. secondary (or side) addressees (among ratified recipients);
 Bystanders (either overhearers or eavesdroppers) (among non-ratified recipients).

And we will add another category to those of Goffman:

 Target (Levinson, 1988: 210 ff) or intended recipient (McCawley, 1984), i.e.

the person for whom the utterance is really intended.

The target does not always coincide with the addressee. When there is a dis-
crepancy between these two types of recipients, we speak of a ‘communicational
trope’ (trope communicationnel)24—a phenomenon which is described masterfully by
Marcel Proust in various passages of À la recherche du temps perdu (translated from
the French):

24
See Kerbrat-Orecchioni (1990: 92 ff) on the different forms of this phenomenon (also see Mizzau
(1994) on triangolazione communicativa). I introduced the term ‘trope communicationnel’ in L’implicite
(Paris: Colin, 1986), where I propose an ‘extended theory of the trope’. I then studied this mechanism on
several occasions, in particular as it operates in drama: theatrical discourse can indeed be considered an
immense communicational trope, since the audience is an ‘eavesdropper’ for the characters, but a target
for the author and the actors. Interviews and various types of talk on the media operate to a certain extent
in the same way—cf. Greatbatch (1992: 269–270), who considers the audience as the ‘primary address’ in
news interviews (the term ‘primary’ actually meaning here ‘intended’).
16 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

‘Ah!’ he said, speaking to no one in particular, so as to be heard at the same time by

Madame de Saint-Euverte to whom he was speaking and Madame de Laumes for
whom he was speaking [. . .] (Du côté de chez Swann, Paris: folio, 334; italics added)

‘The duchy of Aumale was in our family for a long time before becoming a part of
the Maison de France’, explained Monsieur de Charlus to Monsieur de Cam-
bremer, in front of an astounded Morel to whom, to tell the truth, the entire dis-
sertation was, if not addressed, then at least intended. (Sodome et Gomorrhe II: 212)

I afforded myself the pleasure of informing her, but did it by addressing the
information to her mother-in-law, as when playing billiards, in order to hit a ball
one plays against the edge of the billiard-table, that Chopin, far from being out
of fashion, was Debussy’s favorite musician. (ibid.: 212)

This indirect strategy of address is applied with varying degrees of ‘audacity’,

depending on the situation. For example, the strategy is more audacious when the real
target is an overhearer than when he or she is a secondary addressee. Additionally, the
communicational trope can be accompanied by a ‘syllepsis’ (or double meaning). It is
obviously compatible with the phenomenon mentioned in Section 3.1, of multi-
plication of the illocutionary values of the utterance depending on its different
addressees. For example, an interview usually begins as follows:

Mr. X, you were born in Paris in 1940. After brilliantly pursuing studies in
philology at the Sorbonne, you were obliged to leave in order to go into mili-
tary service in Algeria [. . .].

This opening sequence is apparently addressed to the interviewee, here a ‘knowing

participant’, for whom its value is at most a request for confirmation. But its main
target is clearly the listener, in relation to whom it has a stronger value, i.e. that of
an informative statement.

6.2.2. Different types of roles

Participation statuses can be called interlocutive roles, but the functioning of an
interaction also involves other types of roles as well: for example, interactional roles,
which are based on the script; examples: moderator, interviewer vs. interviewee,
teacher vs. student, salesperson vs. buyer (these roles being, in part, determined by
social and institutional statuses, such as those of a journalist, a professor, a trade-
sperson); and discursive roles, which are based on tasks which are carried out (giving
the floor, answering questions, etc.).

6.2.3. The recipient design principle

Cicero has already said of the orator:

The eloquent man should demonstrate the wisdom which will allow him to
adapt to circumstances and to people. I do indeed think that one should talk
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 17

neither all the time, nor against everyone, nor for everyone, nor to everyone in
the same way. Thus, the man who is capable of adapting his language to what is
appropriate to each case will be said to be eloquent. (Cicéron, l’Orateur:
XXXV–XXXVI, 123; translated from the French)

Conversation analysts express the same idea in their own terms. Following Sacks
et al. (1974: 727), the Recipient Design Principle is ‘‘the most general principle par-
ticularizing conversational interaction.’’ It is valid for all types of listeners and not
only for the addressee: ‘‘audience design informs all levels of a speaker’s linguistic
choice’’ (Bell, 1984: 161). In the same way, Clark (1989, 1992) also notes that when
we speak, we ‘design’ our utterance with all our potential listeners in mind, but we
do not deal with them all in the same way. Special attention should be paid to
one’s addressee, whereas towards overhearers, ‘‘speakers can legitimately choose
among a range of attitudes’’, such as indifference, disclosure, concealment, or
disguisement (1992: 255–256). Paralleling this, the listeners’ responsibilities are
not the same, nor are their abilities and handicaps. This principle has been
illustrated for all types of participants, that is, side participants (Clark and
Carlson, 1982), audiences (Drew, 1992), and overhearers (Schober and Clark, 1989;
Johnson and Roen, 1992).

6.2.4. Fuzziness and graduality of categories

Attribution of a given status to a participant can pose problems for the analyst,
as it can for the people who are themselves involved in the interaction. The fuzzi-
ness of the markers also brings about the possibility of misunderstandings, invo-
luntary or voluntary. Finally, it can also occur that participants’ statuses can be
seen differently from each other’s points of view. For example, people in dis-
advantaged positions (children, patients, the elderly) may consider themselves to be
ratified participants, but be treated as unratified participants by their partners in
interaction.
All the categories which go to make up the participation framework are gradual,
denoting different degrees of participation, of ratification, of address, etc. In addi-
tion, constant changes in format and in footing can be observed throughout inter-
action. One ‘slides’ from one conversation to the other in gatherings, and even
within the same conversation, interlocutive configurations are frequently moved
around, resulting in what Goffman (1981: 135) calls ‘‘structural instability’’. These
continual fluctuations in the framing can take place by means of either gradual
shifting or abrupt changes. Change is complete in cases of ‘splitting-off’, or partial
when ‘subordinate communication’ arises parallel to the ‘dominating communi-
cation’. According to Goffman, there are three cases of partial change: byplay,
which arises among ratified participants,25 crossplay, which takes place between
25
On byplay also see the analysis proposed by M.H. Goodwin (1991) of playful commentaries made in
the course of a narrative episode during family table talk in an American home. Goodwin shows that the
framework is at the same time both undermined and upheld—generally, the main purpose of byplay is not
to disrupt the dominant interaction. The intrusion can be negotiated by the participants by means of
different procedures, described in detail by Goodwin.
18 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

ratified participants and bystanders, and sideplay, which consists of ‘‘hushed words
exchanged entirely among bystanders’’ (1981: 133–134).
It is up to the speaker to attribute a particular participation status to each
member of the reception format, and first of all, to choose one or several main
addressee(s). Such choices are made on the basis of various principles.26 But they
are also highly adaptable, and, above all, negotiable. Goodwin (1981), for example,
shows that when it seems that the targeted addressee is not listening, the speaker
will fall back on another addressee who seems better disposed to listening, but this
‘sliding’ from one addressee to another will sometimes have to be accompanied by
the speaker’s reshaping and redesigning the utterance. Still another case is when a
speaker favours one member of the audience to an excessive extent in a situation
where the address is supposedly collective, and the person who is ‘too often looked
at’ tries to remedy this embarrassing situation by refusing eye-contact with the
speaker; such a strategy is aimed at getting the speaker to distribute address signals
more equitably.
The point of the preceding remarks is that an addressed participant can behave in
such a way as to display a relative lack of involvement, and an unaddressed parti-
cipant can, in contrast, behave in such a way as to display a wish to be treated as an
addressee—and can even manage to get this to happen. Building the participation
format is a fundamentally collaborative process.

6.3. An example

As a conclusion, to illustrate these principles of analysis, and especially the gra-

duality of the categories, let us take the example, studied by Traverso (1997), of
what goes on in a French post office, and, more specifically, of the reception format
configuration at t1, a moment in time when A, an employee, addresses B, a custo-
mer. At t1: B is ratified and is the main addressee.
The other customers who are standing in line at the same counter, without being
really ratified in relation to the interaction in progress, are at the same time more
‘legitimate’ listeners than, for example, a person sitting near us on a bus (i.e. a veri-
table overhearer, who must hide whatever interest he or she may have in our con-
versation). It is in no way prohibited for a person who is standing directly behind B
in the line at the post office to follow the exchange between A and B, if only to verify
whether his/her turn at the counter is coming up.
The customers in the waiting-line have hierarchical ranking in terms of ratifica-
tion. The first in line is ‘more ratified’ than the last, and can use his or her ‘quasi-
legitimacy’ in order to speed up a transaction which is dragging on too long, for
example, by moving in closer, by leaning on the counter, etc.27 Likewise, the people
26
Such as ‘Preference for the best source’ and ‘Preference in selecting a spokesperson’, in Leonardi and
Viaro’s study (1990) of encounters between doctors and their patients.
27
This behaviour would be impossible now in most French post offices, where a line on the floor marks
off a physical boundary for the exchange taking place at the counter, in order to safeguard the con-
fidentiality of the interaction in progress (since many French residents also do their banking at the post
office).
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 19

standing in the same line as B are more ratified than those who are standing in lines
at the other counters (although the people in the other lines may nonetheless be
taking an interest in the goings-on in front of another counter, if only to change
lines in case their own line stops moving).
The other postal employees behind counters are more ratified than the waiting
customers. They are part of the staff and can, if necessary, be ‘called to the rescue’.
They are therefore legitimate listeners. Among them, C, an on-the-job trainee, has
been given a particular status, close to that of a secondary addressee (as he is more
or less supposed to observe everything that goes on). Other postal workers who
happen to be present in the post office are also legitimate listeners, but they become
ratified participants less easily than the employees actually behind the counters who
are specifically in charge of customer contact.
This post office participation framework constantly changes, not only because
the main addressee at t1 moment in time becomes the current speaker at t2
moment in time, but also owing to perpetual movements within the reception for-
mat, such as ‘broadening’, ‘reduction’, ‘restratification’, and ‘reorientation’ (cf.
Traverso, 1997).
So this is a complex polylogal situation. Observed as it unfolds, it is, in fact,
like a ‘crossroads of interactions’, where interactions between A and various
successive customers, and interactions between A and other postal employees
closely intertwine—and this description is valid only if analysis is limited to
what takes place among the people who are present within the four walls of
the post office. . . But this notion of an ‘interaction behind closed doors’ is
really an artifice, as Latour (1994: 590) reminds us, for, in fact, all interactions
are infinitely open ‘‘to other elements, to other times, to other places, to other
participants’’:

It is said, without looking too closely, that we are interacting with each other
face-to-face. To be sure, but the garment we are wearing comes from some-
where else; the words we are using were not designed for the situation; the
walls we are leaning against were designed by an architect for a client and
built by construction workers, all absent now, although their actions con-
tinue to make themselves felt. Even the person whom we are addressing
comes from a background which goes far beyond the framework of our
relationship. [. . .] If we wanted to draw a spatio-temporal ‘map’ of all that is
to be found in an interaction, and if we wanted to make a list of all those
who are participating in one way or another, we would not see a clearly-
outlined frame, but instead a very disheveled intertwining network implying
an untold number of extremely diverse dates, places and people. (1994: 590;
translated from the French)

This is a salutary reminder of the over-simplistic nature of the classical perspective

on interaction, as well as of our present inability to simultaneously integrate micro
approaches (concerning the interaction itself) and macro approaches (concerning
background and social aspects).
20 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

7. The studies in this publication

This special issue investigates different types of polylogal situations. Michèle

Grosjean’s article examines the participation framework in a particular type of talk-
at-work produced within the hospital context. The article focuses on shift-change
briefing sessions between teams going off duty and teams coming on duty in different
types of hospital wards, revealing gradual shifts in the basic participation structures
of these encounters from straight dilogues between departing and arriving parties to
genuine polylogues, which emerge at certain specific points in time. She emphasizes
the fact that, in such hospital contexts, the functioning of the participation frame-
work can only be described in relation to the professional statuses of the respective
participants.
Véronique Traverso’s study concerns a semi-formal meeting of researchers
belonging to the same research group. After having distinguished among global,
local, and ‘macro-local’ levels, she takes an even closer look at the macro-local level,
examining in detail two phenomena peculiar to the polylogue, the ‘crowding’ phe-
nomenon and the ‘splitting-off’ phenomenon, which she describes in relation to the
topical lines followed by the participants.
In the article by Sylvie Bruxelles and Catherine Kerbrat-Orecchioni, the investi-
gation moves towards the interpersonal relationship level, focusing on still another
type of phenomenon which characterizes polylogues: the possibility of establishing
alliances or coalitions. After examining the various procedures which can be used to
form a coalition, the authors make a distinction between two main categories of
coalitions: those which are imposed by the frame of the interaction and ‘emerging’
coalitions. These notions are illustrated by an analysis of two excerpts of data of
quite different types: a discussion among five participants within the setting of a
French radio program, and ‘quadrilogal’ encounters in French notaries’ offices in
the context of divorce settlements.
Finally, Michel Marcoccia’s study specifically analyses how polylogues function
‘on line’ in Internet newsgroups, via the written channel, while implicitly modeling
themselves on face-to-face speech. The newsgroup’s mode of communication
obviously has a great deal of influence on the structure of exchanges and the way
they work. However, at the same time, Marcoccia concludes that this mode of
communication only makes more salient certain problems which are characteristic
of all polylogal exchanges, whose complexity presently defies all attempts of for-
malization.
Such is indeed the leitmotiv which runs through these studies: that of the extreme
complexity and flexibility of polylogal organizations—especially since, unlike dilo-
gues (which are objects possessing a sort of fractal structure: on a different scale, the
parts have the same structure as the whole), polylogues have an organization which
is so mobile and so changeable that observing them at a t1 point in time can never
provide a representative picture of the whole.
Such complexity would be enough to discourage any researcher. Yet, as early as 1967,
Sacks recommended that the functioning of multi-party conversations be ‘‘investigated
in its own terms, and not merely [. . .] as a variant off two-party conversation’’ (Lectures,
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 21

vol. I: 523), adding that if two-party conversations are ‘‘much blander’’, multi-party
conversations ‘‘could be much more interesting’’ (ibid.: 533). With such encourage-
ment, how can one resist the urge to take on this challenge?28

References

Allwright, Dick, Bailey, Kathleen M., 1991. Focus on the Language Classroom: An Introduction to
Classroom Research to Language Teachers. Cambridge University Press, Cambridge.
Aronsson, Karin, 1996. Collaboration in dialogues. In: Verschueren, J. et al. (Eds.), Handbook of
Pragmatics. Benjamins, Amsterdam, pp. 1–12.
Aston, Guy, 1998. Introduction. In: Aston, G. (Ed.), Negotiating Service. CLUEB, Bologna, pp. 1–23.
Austin, J., Paddy M., 1987. The Dark Side of Politeness: A Pragmatic Analysis of Non-cooperative
Communication. PhD Dissertation, University of Canterbury.
Bargiela-Chiappini, Francesca, Harris, Sandra J., 1997. Managing language. The discourse of corporate
meetings. Benjamins, Amsterdam.
Barthomeuf, Jacques, 1991. Asymétrie et apprentissage dans les activités de groupe en classe. In: Russier,
C., Stoffel, H., Véronique, D. (Eds.), Interactions en langue etrangère. Publications de l’Université de
Provence, Aix-en- Provence, pp. 249–258.
Bell, Allan, 1984. Language style as audience design. Language in Society 13, 145–204.
Berrier, Astrid, 1997a. Four-party conversation and gender. Pragmatics 7, 325–366.
Berrier, Astrid, 1997b. S’opposer dans une conversation à quatre: de quelques moyens. Revue québécoise
de linguistique 25 (2), 13–33.
Berrier, Astrid, 1997c. La conversation à quatre: quelques aspects interculturels. In: Lefevre, M.-L., Hily,
M.-A. (Eds.), Les situations plurilingues et leurs enjeux. L’Harmattan, Paris, pp. 59–81.
Bilmes, Jack, 1995. Negotiation and compromise: A microanalysis of discussion in the United States
Federal Trade Commission. In: Firth, A. (Ed.), The Discourse of Negotiation. Studies of Language in
the Workplace. Pergamon, Oxford, pp. 60–81.
Boulima, Jamila, 1999. Negotiated Interaction in Target Language Classroom Discourse. Benjamins,
Amsterdam.
Brown, Penelope, Levinson, Stephen, 1978. Universals in language use: politeness phenomena. In: Goody,
E. (Ed.), Questions and Politeness: Strategies in Social Interaction. Cambridge University Press,
Cambridge, pp. 56–289.
Brown, Penelope, Levinson, Stephen, 1987. Politeness. Some Universals in Language use. Cambridge
University Press, Cambridge.
Clark, Herbert H., 1989. Understanding by addresses and overhearers. Cognitive Psychology 21, 211–232.
Clark, Herbert H., 1992. Arenas of Language use. The University of Chicago Press, Chicago. (chapter 8
with Edward F. Schaeffer, ‘‘Dealing with overhearers’’, 248–274).
Clark, Herbert H., Carlson, Thomas B., 1982. Hearers and speech acts. Language 58, 332–373.

28
After quite a number of other scientists. . . Multi-participant situations have already been much
investigated in the field of sociology (following the work of Georg Simmel), social psychology, and
communication studies (see e.g. Cragan and Wright (1980, 1990) for a critical synthesis of about one
hundred studies carried out in the 1980s on communication in small groups). Linguistic investigations on
this issue are much less numerous—see among others: on everyday conversations: Tannen (1984), M.H.
Goodwin (1991), Lerner (1993), Müller (1995), Delamotte-Legrand (1995), Vincent (1995), Traverso
(1996), Berrier (1997a–c); on various types of meetings: Cuff and Sharrock (1985), Olson et al. (1992),
Larrue and Trognon (1993), Bilmes (1995), Bargiela-Chiappini and Harris (1997); on various talk-at-work
situations: Lacoste (1989, 1995), Drew and Heritage (1992b), Diamond (1996), Traverso (1997), Goodwin
and Goodwin (1997), Grosjean and Lacoste (1999); on classroom interactions: Pica and Doughty (1985),
Wright (1987), Barthomeuf (1991), Allwright and Bailey (1991), Boulima (1999); on patient-therapist
encounters: Leonardi and Viaro (1990); on telephone conferences: Perin and Gensollen (1992).
22 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

Clayman, Steven E., 1992. Footing in the achievement of neutrality: The case of news-interview discourse.
In: P. Drew and J. Heritage, eds., 470–520.
Cragan, John F., Wright, David W., 1980. Small group communication research of the 1970s: a synthesis
and critique. Central States Speech Journal 31, 197–213.
Cuff, E.C., Sharrock, W.W., 1985. Meetings. In: van Dijk, T.A. (Ed.), Handbook of Discourse Analysis.
Academic Press, London, pp. 149–159.
Dausendschön-Gay, Ulrich, Krafft, Ulrich, 1991. Tâche conversationnelle et organisation du discours. In:
Dausenschön-Gay, U., Gülich, E., Krafft, U. (Eds.), Linguistische Interaktionsanalysen. Niemeyer,
Tübingen, pp. 131–154.
Delamotte-Legrand, Régine, 1995. Polylogue enfantin ou comment dire à neuf. In: Véronique, D., Vion,
R. (Eds.), Modèles de l’interaction verbale. Presses de l’Université de Provence, Aix-en-Provence,
pp. 397–406.
Diamond, Julie, 1996. Status and Power in Verbal Interaction. Benjamins, Amsterdam.
Drew, Paul, 1992. Contested evidence in courtroom cross-examination: The case of a trial for rape. In:
P. Drew J. Heritage, eds., 470-520.
Drew, Paul, Heritage, John, 1992. Analyzing talk at work: An introduction. In: Drew, P., Heritage, J.
(Eds.), pp. 3–65.
Drew, Paul, Heritage, John (Eds.), 1992b. Talk at Work: Interaction in Institutional Settings. Cambridge
University Press, Cambridge.
Edelsky, Carole, 1981. Who’s got the floor? Language in Society 10, 383–421.
Edmondson, Willis, 1981. Spoken Discourse: A Model for Analysis. Longman, London.
Ford, Cecilia E., Fox, Barbara A., Thompson, Sandra A., 1996. Practices in the construction of turns: the
‘TCV’ revisited. Pragmatics 6, 427–454.
Goffman, Erving, 1981. Forms of Talk. Blackwell, Oxford.
Goodwin, Charles, 1981. Conversational Organization. Academic Press, New York.
Goodwin, Charles, Goodwin, Marjorie H., 1997. La coopération au travail dans un aéroport. Réseaux 85,
129–162.
Goodwin, Marjorie H., 1991. Byplay: Participant structure and framing of collaborative collusion. In:
Conein, B., de Fornel, M., Quéré, L. (Eds.), Les formes de la conversation. CNET, Paris, pp. 155–180.
Greatbatch, David, 1992. On the management of disagreement between news interviewees. In: Drew, P.,
Heritage, J. (Eds.), pp. 268–301.
Grosjean, Michèle, Lacoste, Michèle, 1999. Communication et intelligence collective. Le travail à
l’hôpital. PUF, Paris.
Grosjean, Michèle, Traverso, Véronique, 1998. Les cadres participatifs dans les polylogues. In: Cabasino,
F. (Ed.), Du dialogue au polylogue. CISU, Rome, pp. 51–66.
Heritage, John, 1985. Analyzing news interviews: aspects of the production of talk for an overhearing
audience. In: Van Dijk, T. (Ed.), Handbook of Discourse Analysis. Academic Press, London, pp. 95–
117.
Hymes, Dell H., 1974. Foundations of Sociolinguistics. University of Pennsylvania Press, Philadelphia.
Jeanneret, Thérèse, 1991. Fabrication du texte conversationnel et conversation pluri-locuteurs. Cahiers de
Linguistique Française 12, 83–102.
Jeanneret, Thérèse, 1995. Conversations pluri-locuteurs et co-énonciation. In: Véronique, D., Vion, R.
(Eds.), Modèles de l’interaction verbale. Publications de l’Université de Provence, Aix-en-Provence,
pp. 379–390.
Jeanneret, Thérèse, 1999. La co-énonciation en français. Peter Lang, Bern.
Johnson, Donna H., Roen, Duane H., 1992. Complimenting and involvement in peer reviews: Gender
variation. Language in Society 21, 27–57.
Kerbrat-Orecchioni, Catherine, 1990-1992-1994. Les interactions verbales, 3 Vol. A. Colin, Paris.
Kerbrat-Orecchioni, Catherine, 1997. A multilevel approach in the study of talk-in-interaction.
Pragmatics 7, 1–20.
Kerbrat-Orecchioni, Catherine, Plantin, Christian (Eds.), 1995. Le trilogue. PUL, Lyon.
Lacoste, Michèle, 1989. Parole plurielle et prise de décision. In: Le parler frais d’Erving Goffman. Minuit,
Paris. Pp. 257–273.
 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24 23

Lacoste, Michèle, 1995. Paroles d’action sur un chantier. In: Véronique, D., Vion, R. (Eds.), Des savoir-
faire communicationnels. Publications de l’Université de Provence, Aix-en-Provence, pp. 451–463.
Larrue, Janine, Trognon, Alain, 1993. Organization of turn-taking and mechanisms for turn-taking
repairs in a chaired meeting. Journal of Pragmatics 19, 177–196.
Latour, Bruno, 1994. Une sociologie sans objet? Remarques sur l’interobjectivité. Sociologie du travail 4,
587–607.
Leonardi, Paolo, Viaro, Maurizio, 1990. Conversazione e terapia. L’intervista circolare. Rafaello Contina,
Milano.
Lerner, Gene H., 1993. Collectivities in action: Establishing the relevance of conjoined participation in
conversation. Text 13, 213–245.
Lerner, Gene H., 1996. Finding ‘face’ in the preference structure of talk-in-interaction. Social Psychology
Quarterly 59, 303–321.
Levinson, Stephen C., 1988. Putting linguistics on a proper footing: Explorations in Goffman’s concepts
of participation. In: Drew, P., Wootton, A. (Eds.), Erving Goffman: Exploring the Interaction Order.
Polity Press, Cambridge, pp. 161–227.
Marcoccia, Michel, 1995. Les interviews de couples. Réflexions sur le rôle de porte-parole. In: Kerbrat-
Orecchioni, C., Plantin, C. (Eds.), pp. 81–107.
McCawley, James D., 1984. Speech acts and Goffman’s participant roles. In: Proceedings of the First
Eastern States Conference on Linguistics, Ohio State University Press, Columbus. pp. 261–274.
Mizzau, Marina, 1994. Il terzo commodo. Rassegna Italiana di Sociologia XXXV 189–203.
Müller, Frank E., 1995. Interaction et syntaxe. Structures de participation et structures syntaxiques dans
la conversation à plusieurs participants. In: Véronique, D., Vion, R. (Eds.), Modèles de l’interaction
verbale. Publications de l’Université de Provence, Aix-en-Provence, pp. 331–343.
Müller, Frank, 1997. Review of Paolo Leonardi and Maurizio Viaro, 1990: Conversazione et terapia.
L’intervista circolare. Journal of Pragmatics 28, 383–412.
O’Connell, Daniel C., Kowal, Sabine, Kaltenbacher, Erica, 1990. Turn-taking: a critical analysis of the
research tradition. Journal of Pycholinguistic Research 19, 345–373.
Olson, Gary M., Olson, Judith S., Carter, Mark R., Storrosten, Marianne, 1992. Small group design
meetings: An analysis of collaboration. Human–Computer Interaction 7, 347–374.
Périn, Pascal, Gensollen, Michel, 1992. La communication plurielle. L’interaction dans les télé-
conférences. La Documentation Française, Paris.
Power, R.J.D., Dal Martello, M.F., 1986. Some criticisms of Sacks, Schegloff, and Jefferson on turn
taking. Semiotica 58, 29–40.
Raffler-Engel, Walburga von, 1983. Nonverbal Behaviour in the Career Interview. Benjamins,
Amsterdam.
Roulet, Eddy, 1981. Échanges, interventions et actes de langage dans la structure de la conversation.
Études de Linguistique Appliquée 44, 7–39.
Roulet, Eddy, 1992. On the structure of conversation as negotiation. In: Searle, J.R. et al. (Eds.), (On)
Searle on Conversation. Benjamins, Amsterdam, pp. 91–99.
Sacks, Harvey, 1992. Lectures on conversation. Blackwell, Oxford.
Sacks, Harvey, Schegloff, Emanuel A., Jefferson, Gail, 1978. A simplest systematics for the organization
of turn-taking in conversation. In: Schenkein, J. (Ed.), Studies in the Organization of Conversational
Interaction. Academic Press, New York, pp. 7–55. (First published 1974 in Language 50: 696-735).
Selting, Margaret, 2000. The construction of units in conversational talk. Language in Society 29, 477–
517.
Schank, Roger C., Abelson, Robert P., 1977. Scripts, Plans, Goals, and Understanding. An Inquiry into
Human Knowledge Structure. Erlbaum, Hillsdale.
Schegloff, Emmanuel A., 1995. Parties and talking together: Two ways in which numbers are significant
for talk-in-interaction. In: Haves, P.T., Psathas, G. (Eds.), Situated order. International Institute for
Ethnomethodology and Conversation Analysis and University Press of America, Washington, DC.
Schober, Michaël F., Clark, Herbert H., 1989. Understanding by adressees and overhearers. Cognitive
Psychology 21, 211–232.
24 C. Kerbrat-Orecchioni / Journal of Pragmatics 36 (2004) 1–24

Sinclair, John McH., Coulthard, R. Malcolm, 1975. Towards an analysis of discourse. The English used
by teachers and pupils. Oxford University Press, Oxford.
Tannen, Deborah, 1984. Conversational Style. Analyzing talk among Friends. Ablex, Norwood.
Traverso, Véronique, 1996. La conversation familière. PUL, Lyon.
Traverso, Véronique, 1997. Des échanges à la poste: dilogues, trilogues, polylogue(s)? Cahiers de
Praxématique 28, 57–77.
Vincent, Diane, 1995. Du dialogue au soliloque: des interactions plus ou moins conversationnelles.
Cahiers de Linguistique Française 16, 53–69.
Witko, Agnès, 2000. Une réunion mise en forme par des écrits. Le conseil Municipal. In: Pène, S. (Ed.),
De nouvelles communications dans les organisations. L’Harmattan, Paris.
Wright, Tony, 1987. Roles of Teachers and Learners. Oxford University Press, Oxford.

Catherine Kerbrat-Orecchioni is a Full Professor at Lumière University (Lyon, France). She also holds the
‘‘Linguistics of Interaction’’ chair at the Institut Universitaire de France. She has been a Visiting Professor
in the French Departments of Columbia University (NYC) and the University of Geneva. Her main
interests are pragmatics, discourse analysis, and interaction. She is the author of several books: La Con-
notation, L’Enonciation, L’Implicite, Les Interactions verbales (3 vol.), La Conversation, and Les actes de
langage dans le discours.