Language and Interaction

<DOCINFO
AUTHOR ""
TITLE "Language and Interaction: Discussions with John J. Gumperz"
SUBJECT ""
KEYWORDS ""
SIZE HEIGHT "220"
WIDTH "150"
VOFFSET "4">
Language and Interaction

Discussions with John J. Gumperz
This book features a fascinating and extended focal interview with Professor John J.
Gumperz, who ranges over his long career trajectory and reflects on his scientific
achievements and how they relate to the contemporary linguistic scene. In this way,
the reader is presented with a snapshot introduction to Gumperz’s work in a
contemporary context.
A number of commentaries provide a stimulating and illuminating series of
theoretical and applied encounters with Gumperz’s work from diﬀerent perspectives.
In so doing, they shed new light on Gumperz’s seminal contribution to the study of
language and interaction. In his Response Essay and in a final discussion, Gumperz
clarifies his views on many of the topics discussed in the volume, as well as sharing
with readers his views on some other approaches to language and interaction that are
closely aligned to his own.
Sociolinguistics, the ethnographic approach to language, language and social
interaction, intercultural communication, communicative conventions, contextuali-
zation — these are some of the key terms which Professor John J. Gumperz discusses in
this wide ranging and searching interview about his career as an anthropological linguist
and sociolinguist interested in cultural diversity and intercultural communication.
Contributions by: Aldo di Luzio, Carlo L. Prevignano, Stephen C. Levinson, Paul J.
Thibault, Afzal Ballim, Susan L. Eerdmans, and John J. Gumperz.
John J. Gumperz, Professor Emeritus of Anthropology, Uni-

versity of California, Berkeley, is one of the founders of Socio-
linguistics whose early work on speech communities and on
the relationship of linguistic to social boundaries helped lay
the basis for much current work in the field. Since the 1970s he
has concentrated on a theory and methods of discourse analy-
sis that can account for the intrinsic diversity of today’s com-
municative environments.
His publications include: Language in Social Groups (1962); Ethnography of Commu-
nication (1964) and Directions in Sociolinguistics (1972/2002), both coedited with Dell
Hymes; Discourse Strategies (1982); Language and Social Identity (1982); and Rethink-
ing Linguistic Relativity (1996), coedited with Steven Levinson. He is currently
working on a collection of studies New Ethnographies of Communication (coedited
with Marco Jacquemet); and Language in Social Theory.
Photograph by Paul Bishop with kind permission.

Language and Interaction
Discussions with John J. Gumperz
Edited by
Susan L. Eerdmans
University of Bologna
Carlo L. Prevignano
University of Bologna
Paul J. Thibault
University of Venice and Lingnan University, Hong Kong
John Benjamins Publishing Company

Amsterdam!/!Philadelphia
TM The paper used in this publication meets the minimum requirements
8
of American National Standard for Information Sciences – Permanence

of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data
Language and interaction : discussions with John J. Gumperz / edited by Susan

L. Eerdmans, Carlo L. Prevignano, Paul J. Thibault.
p. cm.
Includes bibliographical references.
1. Gumperz, John Joseph, 1922- 2. Sociolinguistics. 3. Social interac-
tion. I. Eerdmans, Susan. II. Prevignano, Carlo. III. Thibault, Paul J.
P85.G84 L36 2002

306.44-dc21 2002028005
isbn 90!272!2594!X (Eur.) / 1!58811!304!3 (US) (Hb; alk. paper)
© 2003 – John Benjamins B.V.

No part of this book may be reproduced in any form, by print, photoprint, microfilm, or
any other means, without written permission from the publisher.
John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands
John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
Contents
Preface vii
Chapter 1
Presenting John J. Gumperz 1
Aldo di Luzio
Chapter 2
A discussion with John J. Gumperz 7
Carlo L. Prevignano and Aldo di Luzio
Chapter 3
Contextualizing “contextualization cues” 31
Stephen C. Levinson
Chapter 4
Contextualization and social meaning-making practices 41
Paul J. Thibault
Chapter 5
On Gumperz and the minims of interaction 63
Carlo L. Prevignano
Chapter 6
A commentary on a discussion with John J. Gumperz 79
Afzal Ballim
Chapter 7
A review of John J. Gumperz’s current contributions to
Interactional Sociolinguistics 85
Susan L. Eerdmans
Chapter 8
Response essay 105
John J. Gumperz
vi Contents
Chapter 9
Body dynamics, social meaning-making, and scale heterogeneity:
Re-considering contextualization cues and language as mixed-mode
semiosis 127
Paul J. Thibault
Chapter 10
Continuing the discussion with John J. Gumperz 149
Carlo L. Prevignano and Paul J. Thibault
Bio-bibliographical note 163
Index 165
Contents vii
Preface
The aim of the present collection of papers is to highlight some aspects of

John Gumperz’s analysis of human interaction, that which he calls Interac-
tional Sociolinguistic Analysis. This choice has meant that a number of
themes that are central to his work are less focussed on. Our principal con-
cern has been to propose ways in which Gumperz’s research may speak to the
theory and practice of both communication analysis and interaction analysis,
present and future. Moreover, we seek to promote discussion and debate
between the work of Gumperz and other analytical and theoretical ap-
proaches to human interaction.
At the same time, we would like to situate this volume within a by-now-
consolidated tradition of interviews and peer commentary on the research
Wndings of scholars in the human sciences. Herman Parret’s Discussing Lan-
guage (1974) and the journal Behavioral and Brain Sciences come to mind as
signal contributions conceived in this spirit. The present volume is proposed
to the reader with these considerations in mind.
In his interactional sociolinguistic perspective, Gumperz is interested in
the communication ecologies where participants’ interpretive acts and con-
versational inferencing reXect societal power relationships and ideological
processes in the course of human interaction. The role of contextualization
cues, which Gumperz considers to be pure indexicals (Prevignano and di
Luzio, this volume: 9), is crucial here. On the one hand, contextualization cues
are the means whereby social actors retrieve presuppositions in order to make
sense of what they see and hear in interactive encounters; on the other hand,
they interact with symbolic, fully-coded, lexical and grammatical signs in the
processes of constituting speech events.
We shall now brieXy introduce the speciWc contributions to the volume.
Di Luzio’s introductory essay is an overview of Gumperz’s research program
since his earliest dialectological study of the German Swabian dialect of a rural
community in Michigan in 1954. Di Luzio examines the main contours of
Gumperz’s development as both interactional sociolinguist and anthropolo-
gist, focussing particularly on the development of Gumperz’s research on the
social motivations for linguistic variation, the interpretative basis of interac-
tional sociolinguistics, and the notion of contextualization cues.
The discussion between John Gumperz, Carlo Prevignano and Aldo di
viii Preface
Luzio is the center piece of this volume. It provides a forum in which Gumperz
elaborates on and explains many of the central tenets of his approach. In
particular, the discussion examines the ethnographic basis of Gumperz’s re-
search in sociolinguistics, the need to distinguish linguistic forms from the
communicative practices in which these are embedded and in which they have
their meaning. This is a constant theme throughout this discussion and it is
refracted through an interesting series of reXections on intercultural commu-
nication, social action, communicative conventions, and contextualization.
Levinson’s essay draws on the author’s early association with Gumperz as
a Ph.D. student in Berkeley. Levinson develops an internalist approach to
contextualization cues, arguing that utterances specify their own contexts of
interpretation in often very implicit ways, through the use of a range of verbal
and non-verbal resources. Context in this reading is not something which is
externally imposed on utterances. Levinson’s essay shows how the explicit
propositional meaning of utterances is often signiWcantly modiWed by their
cross-modal connections with non-verbal and more implicit, topological mo-
dalities of semiosis. This further suggests that many of the implicit aspects of
human communication cut across apparent surface distinctions between
semiotic modalities, and that this may lead one to make overhasty and
superWcial distinctions between various meaning-making resources.
Thibault’s essay is complementary to Levinson’s in this sense. Thibault
argues for an approach which suggests a way of cutting across these apparent
‘surface’ distinctions in an attempt to develop a conceptually uniWed ap-
proach to the question of communicative — meaning-making — practices,
irrespective of the semiotic resources that are deployed. To achieve this, he
argues for a three-way distinction between indexical, intertextual, and meta-
textual meaning-making practices. This distinction is proposed as a means
for analyzing the ways in which agents access and co-ordinate their deploy-
ments of semiotic resources in the activity structure-types and discourse
genres of a given culture.
Prevignano’s contribution sheds light on a diVerent aspect of this problem
with reference to what he deWnes as the “minims of interaction”. In this, he
contrasts Gumperz with Grice, Leech, and other pragmatists, whom Pre-
vignano characterizes as “maximists”. In particular, Prevignano examines
some “interaction minims”, which participants deploy in order to signal to
each other what they are doing in the course of the interaction. Importantly,
such interaction minims are embedded in activity structures and entail inter-
Preface ix
pretive principles that are historical rather than universal in character. The
analyst’s interpretation, insofar as it is a hermeneutical reconstruction of the
participants’ interpretation, is historical in two fundamental senses. First,
participants are embedded in a history of their applications and interpre-
tations of each others’ interactional minims. Secondly, the analyst, too, is
embedded in a history of ethnographic and analytical reconstruction and
understanding. This history itself acts on and aVects our understanding as
both participants in and analysts of the object of study — human interaction.
Ballim approaches the same problem of interpreting utterances in con-
text, though from a diVerent perspective — that of computer-mediated com-
munication from the point of view of applied computational linguistics. A
central problem for Ballim is how utterances are interpreted in the light of the
knowledge, beliefs, desires, and intentions of discourse participants. Conse-
quently, he is interested in how discourse participants create models of each
other in the course of interaction. Ballim indicates that the overall system
forms top-down constraints on the ways in which participants make infer-
ences and reason about each other in the course of interaction. His discussion
is informed by his research on the use of language as a communicative me-
dium between humans and computers.
Eerdmans provides a critical account of two exemplar case studies con-
ducted by Gumperz in the area of intercultural communication. She shows
how participants involved in inter-ethnic communication interpret and nego-
tiate their interlocutors’ contributions by embedding what they see and hear
in interpretative frames which may conXict with or misconstrue those of the
interlocutor. Interactional sociolinguistics, with its hermeneutic basis, can
provide a powerful tool for explaining such failures on the basis of its ex-
panded understanding of the interaction between the internal dynamics of
agents — their structures of understanding and inferencing — and the larger-
scale dynamics of the unfolding activity-structure, along with its relations to
the wider social and cultural contexts at play in the interaction. In this way, we
begin to see some meaning-based answers as to why interactants act and
behave in the ways they do, and how misunderstanding and communication
failure occur. Eerdmans concludes her discussion with some comments on the
usefulness of interactional sociolinguistics for both teachers and learners of a
second language. In particular, she draws attention to the practical and theo-
retical insights which a critical awareness of the impact of diVerent cultural
backgrounds can have on teaching and learning contexts.
x Preface
In his Response essay, Gumperz ranges over his career trajectory as socio-
linguist before clarifying and expanding on a number of points raised in the
individual contributions as he responds to these. In particular, he makes
further pertinent observations on how his current thinking about language
and interaction builds on his earlier ethnographic Weld studies. In so doing, he
also clariWes the ways in which linguistic and cultural diversity and sociocul-
tural boundaries are represented in and shape linguistic interaction.
In a second essay, Thibault gives voice to some further developments of
the theoretical issues raised by the individual contributors. In particular,
Thibault returns to the notion of action and interaction as a unifying principle
for the analysis of the multimodal co-deployment of semiotic resources. He
argues for the embeddedness of action in higher-scalar ecosocial environ-
ments which interact with the lower-scalar embodied dynamics of the agents
who participate in discursive interaction. He also explores the relationship
between the notions of fully-coded message content and contextualization
cues, as discussed by both Gumperz and Levinson. In particular, he suggests
how these notions can be brought together in a more theoretically-uniWed
framework by considering the typological and topological dimensions of
language as a form of mixed-mode semiosis.
A Wnal discussion between Gumperz, Prevignano and Thibault rounds oV
the discussions which have taken place throughout this volume. In this way,
Professor Gumperz provides further clariWcation of a number of key issues in
interactional sociolinguistics, the role of inferential processes in interpreters’
understandings of each other’s meanings in interactional events, and the
importance of these insights for understanding the contribution of cross-
cultural factors in communicatively-diverse environments. Gumperz also
provides further comments on the centrality of ideological processes in hu-
man interaction, as well as sharing with his readers his views on a number of
other approaches with similar aims to interactional sociolinguistics. A short
bio-bibliographical note detailing the most salient moments in Professor
Gumperz’s career concludes the volume.
The present is a revised edition of a volume which was issued as a pre-

publication by Beta Press (Lausanne CH, 1997) with the title Discussing Com-
munication Analysis 1: John J. Gumperz. This Preface, as well as Chapters 8, 9
and 10, have been specially written for this new edition.
Preface xi
Bertie Kaal of John Benjamins provided us with constant editorial advice

and support from the outset. To Bertie Kaal, all our thanks and deepest
appreciation.
To conclude, we would like to express our gratitude to John Gumperz for
his interest, patience, and generosity throughout all the stages of preparation
of this volume, which we wish to oVer him on the occasion of his 80th
birthday, 2002.
Susan L. Eerdmans
Carlo L. Prevignano
Paul J. Thibault
xii Contents
Chapter 1
Presenting John J. Gumperz
Aldo di Luzio
As a linguist and anthropologist, John Gumperz has successfully combined

and integrated diverse yet complementary issues: theoretical construction
with empirical research, linguistic with sociocultural facts, cognitive con-
straints with communicative ends, text with context, discourse with ethnogra-
phy, analyst interpretation with member interpretation. In this way, he has
developed a new method of considering and explaining intra- and intercul-
tural linguistic processes in multilingual and multicultural communities in
three continents.
Gumperz dealt with languages and cultures in contact in 1954 with his
Ph.D. dissertation at the University of Michigan. The topic was the German
Swabian dialect of a rural community of the third generation of immigrants
in Washtenaw County, Michigan. In this study, Gumperz showed something
very new and innovative for the dialectology of the time, which was still
strictly diachronically and geographically oriented. He was able to demon-
strate that the actual distribution of dialectal variables was a consequence
of social and religious groupings after the settlement in the USA in the
19th century.
The connection or functional relation between the type of group
construction and that of language shift and variation represents a recur-
rent topic in Gumperz’s subsequent research in very diVerent kinds of
speech communities.
Thus, in the second half of the 1950s, in his two years’ Weldwork together
with an interdisciplinary research team in a North Indian village with strict
caste diVerentiations, Gumperz again dealt with the very important question
of the relation between the diversity of sociocultural groups and language
variation. Contrary to what one would have expected, he proved that the
distribution of linguistic varieties or variables in that North Indian commu-
2 Language and interaction
nity did not reXect either caste diVerentiation or other social factors, such as,
for example, income, educational level, or social status. On the contrary, and
as a decisive factor for the distribution of the linguistic variables, norms were
operating which governed and constrained the social quality of interpersonal
communication. In his analysis of code-switching between Hindi and Punjabi
in Delhi, he made it clear that change of code represents a style which depends
on social norms and may express solidarity and group membership. Accentu-
ating the fact that code-switching is not only a grammatical but also and
especially a discursive process, he showed that this could not be analyzed with
traditional linguistic methods. His analysis of the mutual adaptation of Kan-
nada (a Dravidian language), Marathi and Urdu (two Indo-Aryan languages)
in interactions among their speakers in a multilingual community in the
region of Mahrastra represents a milestone for the analysis of creolization and
language convergence studies. In his analysis, he made clear that under the
inXuence of very frequent code-switching and consequent simpliWcation and
reciprocal assimilation, especially at the syntactic level, the local dialects of the
three typologically very diVerent languages become creole-like varieties of the
same language. They present a similar linear structure and can be reciprocally
translated word for word. What in earlier structuralist and psycholinguistic
approaches was analyzed as interference is now interpreted by Gumperz as
socially and communicatively aVected convergence.
Some very important results of Gumperz’s earlier research are that (i)
diVusion and limitation of linguistic variables are determined and displayed
in interaction; (ii) speakers’ perceptions or deWnitions of language equiva-
lence or diversity do not depend on genetic aYliation; (iii) speech commu-
nities are not linguistically homogeneous; (iv) linguistic variation and
alternation are not random or arbitrary, but communicatively functional and
meaningful. These earlier achievements already evidenced new directions in
sociolinguistics. Gumperz showed how multilingualism or multidialectalism,
as well as style variation, with their dialectical and dialogical aspects, represent
a prototypical paradigm or condition of human communication.
Gumperz’s conception, which strongly contrasts with the traditional and
structuralist one of idealized and homogeneous languages and speech com-
munities, is strictly bound to his new concepts of speech community, lan-
guage repertoire and sociolinguistic co-occurrence rules. Members of the
same speech community do not necessarily speak the same language variety.
The only requirement is that the speakers have in common a linguistic reper-
toire, that is, the whole of the linguistic resources and communicative strate-
gies at their disposal and the rules for using them in communication. These
are conceived as co-occurrence rules or implications, which determine at the
paradigmatic and the syntagmatic levels the choice of units from the reper-
toire in building utterances; they also deWne the stylistic proWle of the discur-
sive units according to the communicative goals of the interaction. In this way,
the apories derived from the dichotomy between langue and parole, compe-
tence and performance, are resolved.
For Gumperz as well as for Hymes, with whom he collaborated in the
1960s in order to deWne some aspects of his approach to the ethnography of
communication, linguistic activity is always situated in a dialogical or interac-
tional context and is revealed within the ethnographic context of the speech
community of which the interactants are members. However, Gumperz tends
to consider the communicative units in which communication is enacted as
types of speech activities or communicative genres rather than speech events.
The interactional sociolinguistics which Gumperz developed in the 1970s
implies a new hermeneutical orientation. It is critically opposed to correla-
tional methods because these are external to discourse and cannot explain the
relation between linguistic facts and social structure. Interactional socio-
linguistics may be seen as a sociolinguistic analysis of dialogue and of the
social meanings displayed in the dialogical interaction against the background
of the ethnography of the sociocultural context of the interactants. Regulari-
ties of linguistic interactions are no longer considered as reXecting inde-
pendent social norms; on the contrary, these norms are considered as
interactively (re)produced in the communicative situation. The task of the
interactional sociolinguistic analysis of determinate communicative situa-
tions is to reconstruct the interactive methods by which speakers use discur-
sive means from their sociolinguistic knowledge (for example, language
choice, code-switching and stylistic variation according to determined co-
occurrence rules) for the production and interpretation of the social meaning
intended. An illuminating example of an interactional sociolinguistic recon-
struction of how, in a determinate sociocultural situation, discourse is struc-
tured and social meaning is produced, is already oVered by Gumperz’s study
of code-switching in the small speech community of Hemnesberget in north-
ern Norway.
A very important element of interactional sociolinguistics is a generalized
theory of contextualization and interpretation in communicative interactions.
According to this theory, the interpretation of meaning is interactively negoti-

ated and constituted, taking into account the knowledge of context which the
interactants have at their disposal. In this way, the Lebenswelt enters into
the discourse.
In order to constitute meanings, speakers activate interpretative frames or
schemata from their experience and from their grammatical, lexical and prag-
matical knowledge. The enacting of these schemata is called contextualization.
Contextualization is thus the process by means of which speakers relate
what is said in an interaction to the context of the background knowledge
which is presupposed. It is a necessary element of communication. Conversa-
tional interpretation or inference goes beyond the decodiWcation of lexical
referential meaning. Presuppositions of context are activated and indicated by
means of contextualization cues, such as, for example, intonation, rhythm,
tempo, loudness, gestures, language or register choice and code-switching,
etc.. They are all means whereby intended meaning is signaled. Contextua-
lization cues are produced and organized according to cultural conventions.
In modern multiethnic society, in spite of eVective referential transfer of
information, a communicatively adequate contextualization of discourse can
fail because of diVerent cultural conventions.
The negotiation of meaning necessarily presupposes assessment and un-
derstanding of the culture and communicative habits of the interlocutors. The
assessment of the meaning in the ongoing interaction is also culturally condi-
tioned. The inXuence of language on culture and of culture on language are for
Gumperz bi-directional. Culture is strictly interwoven with language and is
displayed in the typiWed and conventionalized ways of signaling speech activi-
ties through contextualization cues in interaction.
Thus, for example, Gumperz has shown in one of his studies how, in the
intercourse between Hindi English speakers and native English speakers,
diVerent prosodic and rhetorical conventions lead to problems in the inter-
pretation of the meaning. According to Gumperz, the communicative diY-
culties do not depend on lack of grammatical or lexical competence, but on
diVerences in the rhetorical styles which are learned and used in the diVerent
networks of interpersonal relations.
Problems of intercultural understanding are, according to Gumperz,
much more serious, and put the non-dominant interactants of minority
groups at a disadvantage in institutional settings and interactions, e.g., in
counseling and job interviews, public debates and discussions, formal hear-
ings, law proceedings, formal interactions in the work place or at school. In

many studies in which Gumperz analyzed such situations, he has shown how
the diVerent ways in which interactants realize speech activities, leading to
communicative failure, mostly depend on (i) diVerent cultural assumptions
about the situation and the choice of appropriate speech behaviour; (ii) the
diVerent ways of structuring and developing information and argumentation;
(iii) the diVerent ways in which contextualization cues are used and made
functional. From the study of how communication broke down in particular
cases, he was able to gain new insights into the communicative rules and
forms of sociocultural knowledge which in general govern communication
and constrain the forms of its realization.
Intercultural communication, according to Gumperz, is not conWned to
interethnic encounters, but is prototypical for modern industrialized and
bureaucratized western societies when diVerent subgroups live and interact
with one another in large urban agglomerations. In many institutional en-
counters, one very often has to speak with people outside one’s own group in
order to defend one’s own needs and rights and to win personal and social
control. In such situations, in general, two diVerent forms of speaking and
communicative strategies contrast with one another. These denote and con-
note not only or not necessarily social identity and group membership, but
also symbolize diVerent attitudes and support commitment to diVerent sys-
tems of values. The use of in-group symbols in such situations contrasts with
the need to control access to important resources. From this situation, diVer-
ences in the assessment and interpretation of the rhetorical strategies em-
ployed by the other interactants may follow. These can negatively aVect the
result of the interaction with representatives of institutions or majority groups
for the interactant who has failed to master the appropriate rhetorical forms of
communication. Gumperz shows how power asymmetry is hidden under the
surface of equality, since on the assessment side, the performers are people
who know what to expect in the interaction as well as the rules of their
rhetorical attainment, but who do not make them explicit in such interactions.
Starting from “a microsociological point of view”, Bourdieu has advised
us not to forget that “social structure” is present in every interaction and that
communicative intercourse symbolizes and enacts power structure. Gumperz
has made manifest and accountable the realization of these power structures.
He has empirically reconstructed how language as social behavior produces
and reproduces power structures in the course of interaction. He draws our
attention to the fact that discrimination in interaction has a linguistic dimen-

sion. In this case, it is not important how communication is grammatically
structured, but in particular what it does socially and how it is socially gov-
erned and constrained.
Gumperz’s interactional sociolinguistics has crucially contributed to the
recognition of the ways in which language (or speech) and society (or culture),
the cognitive and communicative aspects of language, the theory and praxis of
both speakers and analysts, are strictly interwoven. He has also shown us how
to account for the links between these.
The methods of analysis of communicative or dialogical interactions
which are based on Gumperz’s interactional sociolinguistics are hermeneuti-
cal and necessarily imply ethnographic analysis and the inclusion of the
situational and ethnographic context which is reXexively reproduced in the
interaction. The criteria used for the interpretation of the social meaning
intended and interpreted in communicative interactions are not a prerogative
of the analyst, but are based on those of the participants in the interaction. The
analyst’s interpretation is a reconstruction of the participants’ interpretation.
The assessment of the analyst who has to reconstruct, that is, to know and
understand, the ethnographic context as if he were a participant member, is a
reconstruction of the participants’ assessments and may possibly be con-
Wrmed by them. Such assessment is based, as Vico would say, on sensus
communis, or on judgement without reXection. Gumperz’s method and
his factual criteria for the analysis of communicative interaction remind us
of Vico’s hermeneutical method and the criteria he stated in his Scienza
Nuova:verum est factum and verum et factum convertuntur.
Chapter 2
A discussion with John J. Gumperz1
Carlo L. Prevignano and Aldo di Luzio
Carlo L. Prevignano: I’d like to begin by asking you about your current
research projects.
John J. Gumperz: My interests are both theoretical and applied. Currently I’m
engaged in empirical research on classroom interaction with eight- and nine-
year-old Spanish/English bilinguals. I use methods of conversational and
discourse analysis to formulate hypotheses about the indirect reasoning pro-
cesses students employ in problem-solving. These processes are for the most
part indirect in that they rely on implicit, taken-for-granted presuppositions
to convey information that adults would expect to be overtly lexicalized. By
formulating explicit, veriWable, analytical assumptions about what these pro-
cesses are, and about how children of that age rely on them and talk about
them with their peers, we hope to help teachers gain better insights into their
students’ ways of dealing with the learning task. Ultimately, we intend to make
training Wlms illustrating such reasoning processes, using audiovisual tech-
niques similar to those employed in the B.B.C. Crosstalk Wlms (Gumperz et al.
1979) to reveal the possible linguistic causes of misunderstandings in intercul-
tural communication.
As with my earlier research, such empirical studies are directly related to
my theoretical interest in showing how detailed analyses of communicative
practices can illuminate basic issues in social theory. I’m about to begin
working on a book which I hope will bring out the import of sociolin-
guistically-oriented pragmatic analysis for our understanding of social pro-
cess. Theorists of many persuasions argue that the social environments in
which we live and act are dialogically constituted. The main question I pose is:
how does verbal communication aVect such dialogic processes and to what
extent do these processes depend on shared linguistic knowledge?
My perspective on verbal communication is grounded in earlier studies
in the ethnography of communication. The key insight here is that ethno-

graphically-based sociolinguistic analysis, if it is to be empirically viable,
must focus on speciWc speech events, deWned as interactively constituted,
culturally-framed encounters, and not attempt to explain talk as directly
reXecting the norms, beliefs and values of communities seen as disembodied,
hypothetically uniform wholes. To look at talk as it occurs in speech events
is to look at communicative practices. Along with others, I claim that such
practices constitute an intermediate and in many ways analytically distinct
level of organization. A sociological equivalent here is Erving GoVman’s “in-
teraction order”, a level of organization which bridges the linguistic and the
social. GoVman’s work on this topic has greatly inXuenced the conversa-
tional analysts’ argument that conversation is separate both from grammar
and from macro-social structures and must be analyzed at the level of “activ-
ity” (their use of this term is diVerent from my own; see below). In my
approach to interaction, I take a position somewhat between that of Erving
GoVman and Harold GarWnkel. The former looked at encounters from an
ethologist’s perspective, while the latter was concerned with the interpretive
processes that make interaction work.
The current trend in my work began with the papers in the 1982 book
Discourse Strategies (Gumperz 1982). Apart from its empirical focus on inter-
ethnic and intercultural communication, this book can also be seen as a Wrst
attempt to explore the role of typiWed communicative practices in interaction,
what levels of linguistic signaling they reXect, how they relate to speakers’
communicative and social background and how they aVect interactive out-
comes in key encounters. It is the focus on the interactive and therefore social
import of the Wne details of verbal communication that distinguishes my work
from others. One of my main concerns is with how we can analyze communi-
cative practices in such a way as to account for participants’ ability to create
and maintain communicative involvement and to achieve their communica-
tive ends.
By background I am a linguist trained in the tradition of Saussure, Sapir
and BloomWeld. Structuralism has been severely and, on the whole, convinc-
ingly criticized by Bourdieu and a host of others. Yet, I believe that the
structuralists’ basic insights into linguistic, that is, phonological and syntactic,
competence and their approach to speaking as a partially subconscious pro-
cess, continue to be useful. The problems arise in analyses of everyday talk. In
re-analyzing my ethnographic Weld data on communicative practices for Dis-
course Strategies, I began to realize that Saussurean phonological and gram-

matical structures deWned in terms of Wnite sets of oppositions and truth
condition semantics could not account alone for the relevant dialogic and
discursive facts of everyday talk. These insights are of course not unique.
Along with others in my Weld, I became aware of the semantic importance of
context. Gregory Bateson had long talked about communication being both
context-creating and context-dependent. Conversational analysts provide im-
pressive empirical evidence to show how interpretations shift as part of the
on-going sequential ordering of an interaction. But I argue that sequential
ordering cannot be taken as a structural given. It presupposes active conversa-
tional involvement on the part of speakers, listeners and audience members.
The ability to create and maintain such involvement rests on shared conversa-
tional inferences.
I have proposed the notions of contextualization cues and contextua-
lization processes as a way of accounting for the functioning of linguistic signs
in these inferential processes. Contextualization cues are a class of what prag-
maticians have called “indexical signs”, which serve to retrieve the contextual
presuppositions conversationalists rely on in making sense of what they see
and hear in interactive encounters. They are pure indexicals in that they have
no propositional content. That is, in contrast to other indexicals like pronouns
or discourse markers, they signal only relationally and cannot be assigned
context-free lexical meanings. Yet they play a major role in transforming what
linguists refer to as “discursive structures” into goal-oriented forms of action.
A main aim of my current work on discourse and conversation is to show how
indexical signs, including prosody, code- and style-switching, and formulaic
expressions, interact with symbolic (i.e. grammatical and lexical) signs, se-
quential ordering of exchanges, cultural and other relevant background
knowledge to constitute social action.
This, in brief, is my long-term research program. In my current writing, I
am attempting to integrate the various strains of thought I have alluded to
above, together with my empirical Wndings on urban communicative prac-
tices, into the outline of a coherent theoretical framework. Sociolinguistic
explanation, if it is to be relevant to today’s concerns, cannot implicitly accept
traditional categories of language, culture and society. I believe that interac-
tion at the level of discursive or “communicative practices”, as Hanks (1996)
calls them in his important book by that title, must be seen as separate from
either linguistic or socio-cultural processes. It is constituted by the interplay of
linguistic, social and cultural/ideological forces and governed or constrained

by partly universal and partly locally-speciWc organizational principles. My
argument is that systematic investigation of these principles can provide a
vantage point for an empirically-based reworking of the established traditions
that continue to follow structuralist practices of separating the linguistic from
the social. The recent linguistic anthropologists’ move from a Saussurian to a
more inclusive, broader, Peircian semiotics that distinguishes between sym-
bolic and indexical signs is a Wrst step in that direction. Symbolic signs
communicate via well-known grammatical and lexical rules. Indexical signs
(and among them contextualization cues), on the other hand, communicate
by virtue of direct conventional associations between signs and context, estab-
lished or transmitted through previous communicative experience. Conversa-
tional inferences build on both signaling processes. A major issue in my own
research is to show, without abandoning what we have learned from structur-
alism, how and under what conditions discursive practices work (a) to create
communicative conventions and (b) to aVect interpretation.
C.L.P.: Can you tell us some more about your idea of contextualization cues?
How do you see these verbal and non-verbal constituents in relation to, for
example, their function? Do you assume there is a one-to-one mapping be-
tween cues and cued elements? Because, in my opinion, there is a problem
here. If you can Wnd a sort of one-to-one mapping or relationship, everything
is Wne, but, in pragmatics, there are usually many-to-many relations. Another
problem: you introduced the idea of discourse strategies more than Wfteen
years ago. GarWnkel and others were talking about inference and...
J.J.G.: Inference and sequentiality.
C.L.P.: Inference and procedures. Do you see a diVerence between ethno-
methods or ethnoprocedures and your discourse strategies?
J.J.G.: If we accept that (a) interpretation is always context-dependent and (b)
contextual presuppositions shaping interpretations are themselves subject to
constant change in the course of an interaction, then we cannot expect to Wnd
one-to-one mappings of form to meanings. There are always several possible
interpretations. Contextualization cues, along with other indexical signs,
serve to retrieve the “frames” (in GoVman’s sense of the term) that channel the
interpretive process by “trimming the decision-making tree” and limiting the
range of possible understandings. In talking about the functioning of indexi-
cal signs in interpretation, it becomes necessary to distinguish between mean-

ing in the linguist’s sense of reference, and situated inferences. The latter are
crucial in communicative practice. In everyday talk, situated inferences al-
ways take the form of assessment of what a speaker intends to convey by
means of a message and these are often quite diVerent from propositional
content. For example, if you and Paolo had been talking, and I asked you,
“What did you just do?”, you will not answer, “I made a statement” or “I
performed a speech act.” A more likely answer would be, “I asked him for a
favor” or “I asked him if he was free this evening.” Moreover, even if you had
said, “I asked him a question,” I would not take this as referring to grammati-
cal or speech act categories, but rather as telling me what you wanted from
him. Communicative practices are actions, and conversational inferences are
made by human agents, acting in the real world.
With respect to GarWnkel: in arguing that all communication is inten-
tional and based on inferences, I build on GarWnkel’s notion of “inferential
process”. But GarWnkel is not speciWc as to what he means by the “historical
method” by which members’ interpretive processes can be retrieved, apart
from saying that we need to resort to background information on how and
why a particular inference came about. How do we know what aspects of
background knowledge are relevant at any one time, and is extra-communi-
cative background knowledge enough? We must assume that information
about contextual frames is communicated as part of the process of interact-
ing, and therefore it becomes necessary to be clearer about the speciWcs of
what happens in the interaction as such to assess what is intended. Conver-
sational analysts set out to do this, and their work has brilliantly shown what
can be learned through turn-by-turn sequential analyses. But, as I suggested,
sequential analysis alone cannot account for situated interpretation. It de-
scribes just one of the many indexical processes that aVect inferencing. I’d
like to argue that assessments of communicative intent at any one point in
an exchange take the form of hypotheses that are either conWrmed or reject-
ed in the course of an exchange. That is, I adopt the conversational analysts’
focus on members’ procedures, but apply it to inferencing. The analytical
problem then becomes not just to determine what is meant, but to discover
how interpretive assessments relate to the signaling processes through which
they are negotiated.
But how can we overcome the inherent ambiguity of inferential processes?
In my empirical studies, I have worked out a set of procedures along the
following lines. Analysis begins with turn-by-turn scanning at two levels of

analysis, content and rhythmic organization. The aim is to isolate sequen-
tially-bounded units, marked oV from others in the recorded data by some
degree of thematic coherence and by beginnings and ends detectable through
co-occurring shifts in content, prosody, tempo or other formal markers.
Lectures, ceremonies of various kinds, interviews, that is, named units of the
type normally studied by ethnographers of communication, are instances of
such events. But event sequences can also be isolated in everyday conversa-
tions and other casual encounters, where, for instance, narrative sequences
may alternate or be interspersed with discussion, argument, banter and the
like. In performing this segmentation, we seek to discover natural units of
interaction that contain empirical evidence that conWrms our analyst’s inter-
pretations, evidence against which to test assumptions about what is intended
elsewhere in the sequence. Such event sequences then form the basic units for
the analysis of conversational inferencing. They vary in length from longer
sequences, which in turn can contain many sub-events, to brief three-part
exchanges of move, countermove and conWrmation or disconWrmation.
In phase two of the analysis, events are transcribed. The goal here is to
prepare “interactional texts” by setting down on paper all those perceptual
cues — verbal and nonverbal, segmental and nonsegmental, prosodic, para-
linguistic and other cues — which past and ongoing research shows speakers
and listeners demonstrably rely on as part of the inferential process. This
enables us not only to gain insights into situated understandings, but also to
isolate recurrent form-context relationships and show how they contribute
to interpretation. These relationships can then be studied comparatively
across events, to yield more general hypotheses about members’ contextua-
lization practices.
To return to conversational inference and its role in communicative
practice. Let me give you some concrete examples to show how I view the
process of understanding. Some time ago, while I was driving to the oYce, my
radio was tuned to a classical music station. At the end of the program, the
announcer, a replacement for the regular host who was returning the next day,
signed oV with the following words: “I’ve enjoyed being with you these last
two weeks.” I had not been listening very carefully, but the extra-strong accent
on “you” in a syntactic position where I would have expected an unaccented
pronoun caught my attention. At Wrst the speaker’s words seemed to suggest
that he intended to produce the Wrst part of a formulaic exchange of compli-
ments. But since there was no one else with him on the program, I inferred
that by the way he contextualized his talk, he was indirectly — without putting
it ‘on record’ — implicating the second part, “I hope you have enjoyed listen-
ing to me” (see Gumperz 1996). A second, somewhat more complex example,
comes from my analysis of the cross-examination transcript of the victim in a
rape trial. Counsel: “You knew at the time that the defendant was interested in
you, didn’t you?”. Victim: “He asked me how I’d been... just stuV like that” (see
Gumperz 1995a).
In both cases, I had to search my memory of past communicative experi-
ence to construct a likely scenario or narrative plot that might suggest pos-
sible interpretations. My initial hypothesis in example one conXicted with
what I knew about the radio program, and this triggered a search for a
diVerent, more plausible, scenario. In the second case, I relied on what I
knew about cross-examinations as adversarial proceedings, where the attor-
ney attempts to expose weaknesses in the defendant’s testimony. But while
these general facts tell us something about participants’ motives in their
choice of verbal strategy, we need to turn to what they actually said to under-
stand what they intended to convey. By the words he chose, and by the way
he contextualized his talk, the attorney raised the possibility that defendant
and victim had had a prior relationship. The victim’s move, on the other
hand, positioned as it is immediately after the attorney’s question, implicitly
argues for a diVerent scenario, one where the two were merely casual ac-
quaintances. In this way, she sought to deny and in a sense ward oV the
questioner’s potential attack on her testimony.
I use the term “activity type” or “activity” to refer to the above type of
constructs or “envisionments”, to borrow Fillmore’s term. My claim is that all
interpretation rests on such constructs. Activities are an aspect of GoVmanian
frames and are subject to constant change in the course of the exchange. That
is, they do not apply to events as wholes, they apply to each component move.
I argue that ultimately all interpretation at the level of discursive practice relies
on these constructs.
This view of understanding has some similarity to Fillmore’s notion of
“scene” that he discusses in his work on the semantics of understanding. But
whereas Fillmore is concerned with physical settings, I take more of a social
perspective. I see activities as evoking the actions of actors engaged in strategi-
cally formulating and positioning their moves in order to accomplish com-
municative ends in real-life encounters. In so doing, they rely on their
presuppositions about mutual rights and obligations, as well as on ideologies

of language and individual personalities, to get their message across. This
implies that, in addition to meaning assessment in the established sense, there
are always social relationships that are continuously negotiated and renegoti-
ated by means of the same interpretive processes by which content is assessed.
It is useful to distinguish between two levels of inference in analyses of
interpretive processes: (a) global inferences of what an exchange is about and
what mutual rights and obligations apply, what topics can be brought up, what
is wanted by way of a reply, as well as what can be put into words and what is
to be implied, and (b) local inferences concerning what is intended with any
one move and what is required by way of a response. In this way it becomes
possible to account for changes in frame as a function of the sequential
positioning of moves. Both levels of interpretation involve activities as cogni-
tive constructs: the Wrst is related to what GoVman calls “framing”, while the
second deals with something like the conversational analyst’s “preference
organization”. While contextualization cues assist in retrieving the knowledge
on which activity constructs are based, they do not work in isolation. Interpre-
tation always relies on symbolic, lexical and indexical signs. But as pure
indexicals, contextualization cues are usually produced and interpreted with-
out conscious reXection, and are therefore particularly useful in revealing
frequently unnoticed aspects of the interpretive process that tend to be highly
sensitive to cultural variability.
I’m not claiming of course that these methods solve the problem of
interpretive ambiguity. The aim is to Wnd likely solutions that are plausible in
that they show how component actions cohere in the light of the event as a
whole, be it a three-part string of moves or a longer encounter. This is of
course quite diVerent from assessing the truth or falsity of speciWc interpreta-
tions. The method resembles conversational analytic procedures of recon-
structing the general procedures members employ in formulating speciWc
actions. I diVer from conversational analysts in that my concern is with
situated, on-line interpretation. I want to show both what the most likely
inferences are and how participants arrive at them. In studies of intercultural
and interethnic communication, these methods have been useful in detecting
systematic diVerences in interpretive practices aVecting individuals’ ability to
create and maintain conversational involvement.
Aldo di Luzio: How do you consider coherence? Is coherence dependent on
the type of activity you are engaged in?
J.J.G.: Not exactly. Our assumptions about relevant activities determine our
view of how the interaction coheres.
A.d.L.: By “type of activity”, do you mean “genre”?
J.J.G.: Not quite. I use the term “genre” for another level of pragmatic analysis,
the ideological level. For me genre is not an analytical category, because, as I
understand it, genre, at least in the Bakhtinian sense of the term, is better
treated as an analyst’s, or for that matter also a lay-person’s, concept for
referring to or labeling texts or speech exchanges.
A.d.L.: A mode of speaking about something.
J.J.G.: It’s not just a “mode” of speaking. When we speak of moral discourse as
a speech genre, for example, we are making an ideologically-charged
metapragmatic assessment. That is, we are engaging in a form of talk about
talk. As I said above, I use the notion of activity to refer to conversationalists’
and analysts’ hypotheses about what cognitive processes are involved in un-
derstanding, hypotheses that can then be validated by methods such as those I
have outlined.
C.L.P.: Could you add some remarks about your idea of linguistic convention
and typiWcation?
J.J.G.: Yes, I don’t want to use the term “convention” in the linguist’s sense of
grammatical convention. I use convention as a general term in the lay sense, as
the outcome of a process of typiWcation.
A.d.L.: In the Schutzian sense?
J.J.G.: In the Schutzian sense. For example, Liliana Ionescu-Ruxandoiu this
morning was talking about “pre-sequences”.2 She said certain expressions can
serve as pre-sequences. This reminded me of a Weld-work experience in a
South Asian village where I lived for almost two years. People would come up
to me in the street and say, “Dinner is ready, come on over.” Once, when I
seemed hesitant, the speaker said, “Don’t you want to come? We’ve made
dinner, why don’t you join us?”. I still hesitated. I didn’t know what to do.
Finally, thinking I had perhaps actually been invited and had forgotten about
it and might oVend them if I didn’t come, I went along. Well, it turned out that
I was...
C.L.P.: Invited?
J.J.G.: I was not invited, no. It was just a greeting. An invitation is a very
solemn matter in the village. To invite someone, you send an emissary, a
young man in your household, to give the invitation. And even then, one
doesn’t go to the host’s house until a second emissary comes to call and asks
the guest to come over. This means that a special meal has been prepared and
is now ready. So that to say, “Dinner is ready”, to someone in the street, simply
counts as a form of greeting.
C.L.P.: This is a convention, the outcome of a typiWcation within a commu-
nity.
J.J.G.: One Wnds similar usages all over South Asia. In some parts of the
continent they say, “Have you eaten yet?”, as a way of conveying something
like, “How are you?”. Such conventions arose over time as outcomes of cultur-
ally-speciWc processes of typiWcation. They may reXect a time when people in
farming communities did not have everything they wanted to eat.
C.L.P.: So when you are confronted with this kind of utterance, if you come
from another community, like myself, you take it as an invitation.
J.J.G.: Exactly, as I did. In fact, once I went, and they had to prepare the food.
And I don’t know whether they had enough to feed me. They were not poor
people — it was a wealthy village — but still, you know, the women had to get
to work and make some more food.
C.L.P.: To come back to your notion of contextualization cues. You said that
you are not interested in the propositional content, but in the indexical
function in the Peircian sense, versus the symbolic, is that right?
J.J.G.: Yes, yes.
C.L.P.: So you see cues as orienting people to...
J.J.G.: Certain interpretations.
C.L.P.: And what about the orienting relationship? You have an addressee
who has to be oriented to, who has to be instructed via cues...
J.J.G.: That’s right.
C.L.P.: However, there are diVerent competences, repertoires, but let us as-
sume that you have two people with just the same repertoire or competence.
Faced with the same cue, they are supposed to be oriented in the same way, in
the sense that they are supposed to reach the same cued element ...
J.J.G.: The same inferences. Be able to draw similar inferences, yes.
C.L.P.: But this is not always the case. Given these two people belonging to the
same community, with the same repertoire of cues, are we sure that there are
no cases of ambiguity, contrary to the one-to-one relationship?
J.J.G.: There always are some ambiguities.
C.L.P.: There always are, okay.
J.J.G.: There usually are. Ambiguities always exist. This is why I emphasize
that the basic issue is not whether or not people understand factual informa-
tion, but whether or not participants in an encounter are able to attune to each
other’s interpretive processes. For example, ambiguities and misunderstand-
ings always occur, but you need to be able to repair them. And since conversa-
tional repairs must, for reasons inherent to conversing, rely in large part on
indirectness, repairs always make unusually high demands in the way of
shared inferences. In the cross-examination example I gave before, it is not
that the victim and the attorney were unable to draw similar inferences. On
the contrary, given the way jury trials work, they chose intentionally to rely on
shared interpretive conventions to convey conXicting accounts of what might
have happened.
C.L.P.: During the IADA Round Table, Edda Weigand spoke about coming to
an understanding in dialogue, a sort of Habermasian idea. However, you are
never sure you have reached a total agreement, a convergence of inferences, of
schemata and so on. So misunderstanding is a human condition, and you are
alone, in a certain sense, with your cues, facing the other person. And this is
the reason why I see individuals as a bit like solipsistic entities. They just send
and get cues, and cues about cues, trying, via them, to instruct the other’s
inferences or schema use.
J.J.G.: But we are not really talking about individuals agreeing on what some-
thing means. Nor is it so easy to say when people have the same communica-
tive background. Let me give another example. One of the very best of my
former students, a Nigerian, Niyi Akinnaso, was born in a Nigerian village. He
was the Wrst literate person in the community. His father, a local chief, was
secretary of the farmers’ cooperative and he used to keep accounts using a
local mechanical device, something like an abacus. As a child, Niyi used to
keep records for his father. He went to missionary school in the region and
ultimately to the University, became an instructor there and then came to
Berkeley in the U.S. for his Ph.D. He now has a professorship in the United
States and specializes in issues of literacy.
Niyi and I now move in the same circles and are part of the same academic
community. I talk with him about a range of academic and other issues in
ways that I cannot with many others in the U.S., and would certainly never be
able to talk to his father or anyone else in Nigeria, although Niyi can. Through
participation in similar “networks of relationships” over time, we have been
socialized into similar network-speciWc communicative practices. Although
our backgrounds are about as diVerent as they could be, we share certain
communicative conventions and interpretive practices. It is long-term expo-
sure to similar communicative experience in institutionalized networks of
relationship and not language or community membership as such that lies at
the root of shared culture and shared inferential practices. In most people’s
lives, community membership is of course directly linked to participation in
such networks of relationships, but in our post-industrial worlds, it is less and
less possible to take this for granted.
Apart from cultural sharing in the face of diVerences in background, there
are now more and more cases of people brought up in what by ordinary
criteria counts as the same community, but in whose case surface similarity of
language and background hides deep underlying diVerences.
To go back to what I said about shared interpretations. It is not merely a
question of what something ‘means’. Ultimately, agreement on speciWc inter-
pretation presupposes the ability to negotiate repairs and re-negotiate misun-
derstandings, agree on how parts of an argument cohere, follow thematic
shifts and shifts in presuppositions, that is, share indexical conventions. The
more basic issue is to show how these tasks are accomplished. And it is for this
reason that my analysis puts so much stress on contextualization processes.
C.L.P.: You start from the idea of “metamessages” à la Bateson or Watzlawick,
and you seem to give great importance to these repair/adjustment procedures
via metacues, cues concerning other cues, or metapragmatic cues.
J.J.G.: Actually, the Wrst paper I did on this topic was a paper I gave in Urbino.
C.L.P.: It was published in 1974.
J.J.G.: In the 1974 series, yes (Gumperz 1974). That was my Wrst, but I didn’t
have the theory then. But now...
C.L.P.: As to the metapragmatic activity, I think it is important in order to

survive. If you don’t have metapragmatic strategies ...
J.J.G.: It is not only that you don’t survive, you don’t learn. You cannot
proWt from your own misunderstandings. And this clearly has to do with
verbal ability at the level of communicative practices, because it is on such
practices that our ability to assess and evaluate the signiWcance of what we
perceive rests.
C.L.P.: In psychiatric cases, it’s just these metapragmatic procedures that fail.
You try, you try, and the other...
J.J.G.: Exactly, exactly. I have had students who worked with family psychiatry
transcripts, from the Watzlawick school. And although no complete analysis
was done, it would seem that even in those cases, much could be learned from
close, turn-by-turn analysis of communicative practices.
C.L.P.: Another point. You mentioned GarWnkel and you said, “I am a little bit
in between GarWnkel...”
J.J.G.: And GoVman.
C.L.P.: And you said that from GarWnkel you took the problem of inferential
processes.
J.J.G.: That’s right.
C.L.P.: So let us discuss, if you agree, the problem of dialogue and reasoning,
the reasoning required by dialogue. Because dialogue requires reasoning.
J.J.G.: That’s right, you’re always reasoning. Conversational inference always
involves reasoning in some form. You start — I’ll use the term ‘assessing’ or
‘assessment’ — you start by assessing in order to be able to recall a message.
You always ask yourself both, “What are they saying?” and “What do they
intend?”. Assessments begin at the level of phonetics, and segmental pho-
nemes are of course important — you need to be able to assess whether for
example you are hearing ‘g’ or ‘k’. But rhythmic organization and prosody
may be equally important in making the relevant assessment.
In fact, in my Urbino paper, which formed the basis for the Wrst chapter of
Discourse Strategies, I discussed the example of a graduate student who asked
me after class, speaking in standard-like style, “Can I come to see you?”.
I replied, “Sure, come to the oYce,” whereupon he went on to say in what

seemed like old-fashioned Afro-American English, “Ahma git me a gig.” At
Wrst I wondered why he chose to switch codes in this academic context. He
usually spoke much like everyone else, so the switch must have been inten-
tional. Was he rejecting the ‘academic norm’ of Standard English? It was only
after discussing the incident with several other graduate students familiar with
Afro-American interpretive traditions that I recognized that, by speaking the
way he did, he was signaling that he was no longer addressing himself just to
me, but to anyone who shared his interpretive conventions, that is, mainly —
but not exclusively — the other African-Americans in the group. What he was
trying to convey can roughly be paraphrased as: “I am doing something that
minority people like myself have to do, get support where I can.” In other
words, when his utterance is interpreted in terms of what we know about the
positions of African-Americans in the urban U.S., and in terms of cues such as
the undiphthongized lengthened ‘ah’ instead of ‘I’ and the lengthened vowel
in ‘gig’, as well as the highly-contoured intonation he used, it becomes clear
that he was employing a formulaic expression, which, in the context at hand,
indexically conveyed his message. He was not ‘code-switching’, but using
metapragmatic strategies to convey a message.
In a related example from Discourse Strategies, a young elementary
school student, when asked to read, replied, “Ah cain’t read.” The teaching
aid thought he meant to say that he was not able to read. Examples like this
have often be cited in the literature on classroom learning in support of
assertions that African-American students have more diYculty with literacy
learning than others. But when we discussed this example with a group of
African-American graduates, they pointed out that the expression carried
contoured intonation and, given the expression’s positioning after the tutor’s
question, they would interpret the student as saying essentially, “I don’t want
to do it right now. I want company in reading.” On the basis of experiences
like these, I became alerted to the fact that for those familiar with African-
American conventions, the opposition between contoured and non-con-
toured intonation may be information-carrying. Such uses of prosody are
not only found in African-American speech. Consider the following example
I recently heard in Cambridge from someone talking about King’s College:
“Fellows of King’s are well-known, to fellows of King’s.” The second (itali-
cized) phrase is set oV from the Wrst and therefore foregrounded by lowering
of pitch and volume, and this suggests that “well-known” is restricted so as
to highlight the interpretation that the compliment phrase applies only to
other fellows of King’s, not to the public at large. Understanding, as struc-

turalists have taught us, always relies on selective perceptions based on our
knowledge of oppositions. This is true for indexicals as well as for symbolic
relationships. All interpretive assessments are relational (note the Saussurian
heritage here). They are made with reference to something else, not neces-
sarily directly represented in talk. But nevertheless, as the examples show,
assessing involves reasoning which is intrinsically dialogic in the sense that
positioning within an exchange is crucial.
C.L.P.: Two other questions. What do you think about what could be called
“corpus pragmatics”? And the second question, what do you think about the
notion of “pragmatic creativity”?
J.J.G.: Corpus pragmatics?
C.L.P.: Or corpus linguistics, in this case for pragmatic analysis.

J.J.G.: Okay, we always rely on a corpus, in the sense that to analyze anything
at all in any depth we must prepare written transcripts. My problem with
corpus linguistics is that it treats talk as if it were a literary text. And as to
corpus pragmatics, pragmatics always requires us to take context seriously.
This suggests we cannot base our analyses on a single corpus, we must work
comparatively. We must Wnd ways of systematically contrasting our analysis
with other comparable ones carried out under diVerent but comparable con-
textual conditions. This requires us to address the question of the criteria we
use for determining similarities and diVerences in and across contexts.
A second question that arises, if we take seriously the need for compara-
tive analyses in pragmatics, is the distinction between communicative practice
as a form of action and language as linguistic form. This is an issue that has
only begun to be addressed in current comparative research in linguistic
anthropology, but so far it has not, as far as I know, been systematically
considered in corpus linguistics or, for that matter, in pragmatics.
C.L.P.: When talking about corpus linguistics, I have in mind the rather recent
wave of electronic corpora.
J.J.G.: But you mean ...?
C.L.P.: Sinclair, for example, John Sinclair’s work.
J.J.G.: John Sinclair, yes, and the Lund people.
C.L.P.: For example, yes. But the problem, from your perspective, is how to
record the contexts as well...
J.J.G.: That’s right, exactly.
C.L.P.: That is the question.
J.J.G.: And also, it is not the corpus as such that is the problem. It is a question
of context, as well as of more basic questions of analysis. Take transcription.
We have as yet no generally agreed-upon, universally applicable system of
transcription. Not even something like the international phonetic alphabet.
And transcription, as recent work has shown, must always be related to
analysis. For conversation, the so-called Gail JeVerson transcription system
has come to be widely used throughout the world. But in my terms, it neglects
communicatively signiWcant prosodic and paralinguistic aspects of speech. It
cannot, for example, account for the interpretive import of phonetic variabil-
ity and so on.
There is also no agreement on methods of analysis. Quantitative methods
are widely used, but in the absence of agreement on what is signiWcant and
countable, quantitative methods are limited in value. Therefore, I currently do
not Wnd it productive for the issues that concern me. There has to be a division
of labor.
C.L.P.: Sure. It’s a problem I’ve found, too. When I have a number of cases
before me, I am faced with a set of pairs of text and context. But this is much
more than a corpus of texts only. Instead, when you have an electronic corpus,
you have a set of utterances only, you can enumerate their constituents, their
co-occurrence, and so on.
J.J.G.: There are quite sophisticated methods of computer-based scanning and
retrieval that work on raw data. But we need agreement on what to look for
and what the goals of the analysis are. Most discourse analysts work with
lexicon and clause level grammar and are concerned with issues of structure.
C.L.P.: Okay, it’s possible to carry out all kinds of search procedures, but given
this possibility, in your opinion, what’s the sense of these procedures with
regard to the problem of context?
J.J.G.: Such analyses cannot account for the eVect of context on interpretation.
You always need a prior analysis of context. You need independent, ethno-
graphically-based analysis and information on the relationship between lin-

guistic forms and discursive structures.
C.L.P.: I think that as an analyst, an imperfect analyst, I always have to resist

the temptation of omniscience and to try to reconstruct. But what does it
mean to be an analyst, even if consciously not omniscient, and to re-con-
struct? It depends on the cues which are at your disposal.
J.J.G.: You need more background information than you ordinarily have as an
investigator who simply elicits and transcribes talk. This information is best
collected through ethnographic observation and participation in everyday
routines such as those we relied on in the classroom interaction study I talked
about at the beginning of this interview. It is also useful to make preliminary
analyses and test one’s assumptions about interpretive conventions through
informal discussions of analyzed texts along the lines of the procedures I
discussed in Discourse Strategies. Ideally, such ethnographic procedures
should accompany or precede large-scale text analyses.
C.L.P.: So, can you have a corpus of texts with contexts?
J.J.G.: I don’t know if I would use the term “corpus” here. There are ethno-
graphic studies of communicative practices. Marco Jacquemet’s (1996) book
on the Naples Camorra trial is such a study and should provide an analyst with
at least some background information on discursive practices in the area of
Naples. Another book in my Cambridge series, Linda Young’s Crosstalk and
Culture in Sino-American Communication (1994), is also very useful in show-
ing how the Chinese frame their interactions.
C.L.P.: But if you try to reconstruct, what kind of generalization can you
reach? For instance, there is a very interesting example of yours quoted by
Deborah SchiVrin in her book Approaches to Discourse (1994:100), the ex-
ample of the teacher and the child. When the child replied, “I don’t know,” it
was a sort of cue, as if to say, “Please give me time.”
J.J.G.: “I don’t know”?
C.L.P.: “I don’t know,” in this sense, a sort of hesitation, conveying the wish to
be encouraged. But the teacher took this kind of utterance as meaning: “I
really have no knowledge concerning that.”
J.J.G.: Exactly, yes.
C.L.P.: Okay, this is one case. You were able to reconstruct the dynamics of
this case...
J.J.G.: Only because I happened to have a graduate student with me who was
familiar with the relevant conventions. But one or two instances are only
suYcient for initial hypotheses. After that I had to go on to Wnd other in-
stances of the same phenomenon in order to construct a more general picture.
That is the way ethnography works. It is a matter of observation, hypothesis
formation, hypothesis testing on the basis of further information and then
critiques from locals who are familiar with the situation, and so on.
C.L.P.: Yes. But how can I continue? For example, I have tried to examine a set
of already published examples, already studied by other people.
J.J.G.: That’s very hard.
C.L.P.: And the problem is that you can reach diVerent conclusions, diVerent
hypotheses, even if you are still attached to the same presuppositions and
assumptions, because you can attend to or focus on diVerent subsets of cues
within the same overall set of cues.
J.J.G.: One solution is to have a local assistant who works with you on the
analysis and who will tell you what it is in what he/she hears or perceives that
leads to the interpretation. That gives you information at two levels, content
and form. The more one works with such interpretive analyses, the more
native-like one’s interpretations can become.
C.L.P.: But you can reach diVerent conclusions, given the same inputs or cues,
even if you are a native interpreter. So I think that the enterprise of prag-
maticians is a sort of game whose rules still have to be determined.
J.J.G.: It’s only a game if you share the rules, otherwise it can’t be a game. It’s
the kind of game ethnographers must learn to play. And many of them fail to
do it. The trick is to learn to ask the right kind, i.e. productive, questions.
Questions that lead to answers that give one the feeling that progress is being
made even though many problems remain.
C.L.P.: In your opinion, what kind of generalizations can we reach, starting
from these cases? Because they are single cases.
J.J.G.: Ultimately, large-scale quantitative studies are necessary. But current
studies of this kind tend to work with many unexplicated assumptions. Good,
valid ethnographic information can lead to more sophisticated and informa-

tive quantitative analyses. A basic problem in sociolinguistic research is how
to deWne the domain of our analysis. In the case of standard languages like
English or Italian, the problems are relatively minor. But how do we deWne
what is or is not Romansch or Piedmontese, apart from basing our sampling
on a geographically-bounded region? In discourse analyses of such genres as
classroom lecturing, similar problems arise. Initially, students of classroom
discourse simply asked teachers to lecture on speciWc topics and recorded
what they did. Obviously, that kind of procedure raises problems about the
generalizations one can derive from one’s data. At the other extreme, some
researchers simply turn on the machine and record everything that is said.
This raises other analytical problems. All these are problems that can be
avoided or at least clariWed given basic background information and theoreti-
cal insights into how communicative practices work.
C.L.P.: I think that you adopt a conception which is sociocognitive. Aren’t you
afraid of the mentalistic objection?
J.J.G.: No, because my analyses are based on empirical and I hope replicable
data. I talk about inferences, but these are grounded inferences, based, as I said
before, on two sorts of data, at the level of form and at the level of content, as
well as on background information on interpretive practices, and symbolic
and indexical signs. If it can be empirically shown that the inferential pro-
cesses I postulate also hold true in other comparable contexts and are similarly
grounded in verbal signs, I can make Boolean generalizations about relations
between classes of objects. But I cannot make valid predictions about speciWc
interpretations. Dialogue analysts are right. We are limited in some ways, but
maybe not in the way they say we are.
C.L.P.: As to the second question, what do you think about pragmatic creativ-
ity?
J.J.G.: Pragmatic creativity has to refer to innovation in the context of certain
constraints. It’s akin to creativity in poetry or writing. Writers can be creative
or crazy.
C.L.P.: Craziness is maybe a kind of creativity.
J.J.G.: We often assume that people are “crazy” when we fail to see order or
coherence of some kind in what they say or do. Creativity bends boundaries
without violating our sense of order.
C.L.P.: So, in your sense, pragmatic creativity can be a sort of violation of

conventions?
J.J.G.: Yes, it can be understood in terms of a broader set of conventions.
These apparent violations can be understood in terms of our broader sense of
order.
C.L.P.: Okay, here I think Mr. Grice enters the scene...
J.J.G.: Yes, given a broad interpretation of Gricean inference.
C.L.P.: ...with “violation” and “Xouting”. From your perspective, how is it
possible for conversationalists and analysts facing some inputs, some cues and
so on, to decide whether they are creative or not?
J.J.G.: That always depends on your background and speciWc circumstances.
C.L.P.: I see it as the problem of the existence of unconventionality. You are
always faced with conventions, maybe conventions coming from another set
or system of conventions. You can take them as new things or violations, but
in fact they belong to other sets of conventions. But consider someone who
wants to be unconventional within the same community, that is, within a set
of established conventions. In your opinion, what does it mean to be uncon-
ventional?
J.J.G.: To do something that seems to violate one rule, but that can often be
explained in terms of another, broader principle.
C.L.P.: In your opinion, can we capture violation with procedures, with
strategies?
J.J.G.: Yes. Irony and humour, as Grice points out, often depend on such
violations. I gave an example of irony in my article in Auer and di Luzio’s
book, The Contextualization of Language (Gumperz 1992). Three graduate
students are talking about why they chose the subjects they are studying and
one of them, hearing that the other is going into engineering, exclaims, “We
Asians.” To me as an analyst, the utterance seemed strange and certainly not
on topic. I only found out afterwards, when I inquired, that his reaction to his
friend’s statement was an ironic allusion to a stereotype about Asians, who are
aware of the so-called glass ceiling which keeps them from being promoted to
important managerial positions and therefore prefer to stick to technical
occupations. In a way, the graduate student was being creative in calling his
fellows’ attention to a problem they all face.
C.L.P.: A last question. In your career, have you ever tried to describe a frame,
a script, or something like that?
J.J.G.: Bateson likens framing in communication to framing in paintings.
GoVman talks about how frames aVect the way we perceive an interaction and
how they change in the course of the interaction. For me framing is another
way of talking about the activity level presuppositions that aVect interpreta-
tions at any one point in an exchange. In my studies of South Asians living in
England, I have shown that frequently they perceive and frame encounters
quite diVerently from their native-speaking interlocutors. In one particularly
striking case, an interviewee seemed quite unreasonably to be denying or
contradicting information that the interviewer told him, even though only the
interviewer had Wrst-hand knowledge of the topic. It took us a long time to
discover that the interviewee was simply pleading for sympathy or under-
standing. Based on these and a series of other miscommunications, I was able
to argue that many South Asian speakers have ways of framing formal inter-
view-like encounters which are quite diVerent from those Westerners are
familiar with. A “frame”, as I use the term, can be described as a class of related
presuppositions that guide conduct in certain situations and that are directly
related to choice of topic and verbal strategies. I gave an example of this in my
paper in Mutualities in Dialogue (Gumperz 1995a).
C.L.P.: You said (in Gumperz 1995b) that Foppa, one of the editors of Mutu-
alities in Dialogue, asked you something about your paper.
J.J.G.: He asked me a question, one that I frequently hear: “How do you know
if people understand each other? When do I know that my interpretation is
the right one?”. And I said, “Well, only through context do you have the right
interpretation, and you’ll reach an agreement, but only as the result of a
dialogical process.”
C.L.P.: How can one study the cognitive aspects of this process?
J.J.G.: This is actually what I will talk about tomorrow.3 I will take two
examples. There’s a rape case, a re-analysis of Paul Drew’s example from a
rape case, and I’ll talk about some of the grammar that’s involved there. I’ll talk
about what each person has to know in order to do what they’re doing. And
then I’ll take my example from a job interview to show again that they did not
share that knowledge, that therefore they couldn’t understand, and the more
they tried, the more things went wrong.
C.L.P.: That’s very interesting.
A.d.L.: Just to go back to the notion of genre for a minute: it is not very clear to
me. Is it a metapragmatic category?
J.J.G.: We use the term “genre” to label, that is, to refer to people’s modes of
talking. In a way genre is similar to what some people call speech style. But
genres are always associated with speciWc ideological values. As Hanks and
others have pointed out, by using speciWc genres, we allude to the values that
are associated with them.
A.d.L.: Yes, that’s what I thought.
J.J.G.: But the only thing is that I talk from the speciWc standpoint of some-
body who has either heard about it or analyzed it, and I am aware of what
other people think about it.
A.d.L.: I understand it as a sort of art of organizing activity in a special way, as
Luckmann says, “to solve a communicative problem”.
J.J.G.: Yes, that’s right. We organize activity to solve a communicative prob-
lem, but I would limit it. I would say that we do that in an activity, but genre is
the way we talk about it, and the way it gets transmitted and passed on. And in
fact I would say, when we talk about dialect, for example, Konstanz dialect
versus Kreuzlinger dialect, we’re making a judgement. That is, a judgement at
the level of genre. And what I would argue is that any attempt to delineate the
notion of dialect and style empirically, in terms of some kind of Saussurean or
empirical analysis, is not going to be very successful. You don’t need agree-
ment on the linguistic facts of what a genre is, as long as there’s some sharing
of certain key elements that would be the prototypical judgements that we’re
making, not a systematic analysis, that’s the point.
Notes
1. This discussion took place on March 31, 1995, at the Department of Linguistic and
Oriental Studies, University of Bologna, Italy, on the occasion of the 1995 Bologna Round
Table on Dialogue Analysis: Units, Relations and Strategies beyond the Sentence, organized
by the International Association for Dialogue Analysis (IADA) in honour of its President,
Sorin Stati. Transcribed and edited by Susan Eerdmans. Revised by John J. Gumperz and
received February 1996; revised for the present edition February 2002 [Editors’ note].
2. Gumperz is referring to the paper which Ionescu-Ruxandoiu Wrst presented at the 1995
IADA Round Table as “Pre-sequences as a conversational strategy”, now published as
Ionescu-Ruxandoiu (1997) [Editors’ note].
3. Gumperz concluded the 1995 IADA Round Table with a seminar on his work in
interactional sociolinguistics (see Eerdmans, this volume) [Editors’ note].
References
Eerdmans, Susan L. “A review of John J. Gumperz’s current contributions to Interactional

Sociolinguistics”. This volume.
Gumperz, John J. 1974. The Sociolinguistics of Interpersonal Communication. Urbino,
Italy: Centro Internazionale di Semiotica e di Linguistica. Working Papers and Pre-
publications, 33, serie C.
Gumperz, John J. 1982. Discourse Strategies. Cambridge: Cambridge University Press.
Gumperz, John J. 1992. “Contextualization revisited”. In The Contextualization of Lan-
guage, P. Auer and A. di Luzio (eds.), 39–53. Amsterdam/Philadelphia: John Ben-
jamins.
Gumperz, John J. 1995a. “Mutual inferencing in conversation”. In Mutualities in Dia-
logue, I. Marková, C.F. Graumann and K. Foppa (eds.), 101–123. Cambridge: Cam-
bridge University Press.
Gumperz, John J. 1995b. Personal communication to C.L. Prevignano, Bologna, 31
March.
Gumperz, John J. 1996. “The linguistic and cultural relativity of conversational inference”.
In Rethinking Linguistic Relativity, J.J. Gumperz and S.C. Levinson (eds.), 374–406.
Cambridge: Cambridge University Press.
Gumperz, John J., Jupp, Tom C. and Roberts, Celia. 1979. Crosstalk: A Study of Cross-
Cultural Communication. London: National Centre for Industrial Language Training
in association with the B.B.C..
Hanks, William F. 1996. Language and Communicative Practices. Boulder, CO: Westview
Press.
Ionescu-Ruxandoiu, Liliana. 1997. “Pre-exchanges as a conversational strategy”. In Dia-
logue Analysis: Units, Relations and Strategies beyond the Sentence, E. Weigand (ed.),
121–128. Tübingen: Niemeyer.
Jacquemet, Marco. 1996. Credibility in Court: Communicative Practices in the Camorra
Trials. Cambridge: Cambridge University Press.
SchiVrin, Deborah. 1994. Approaches to Discourse. Oxford: Blackwell.
Young, Linda W.L. 1994. Crosstalk and Culture in Sino-American Communication. Cam-
bridge: Cambridge University Press.
Chapter 3
Contextualizing “contextualization cues”
Stephen C. Levinson
I would like to focus my remarks on John Gumperz’s notion of a contextua-

lization cue, which I think is central to his current program. However, before I
begin, I would like to make some more general remarks on that program and
its place both in Gumperz’s overall oeuvre and in discourse studies generally.
As di Luzio sketches in his contribution to this volume, Gumperz had a
previous incarnation before becoming the founding father of interactional
sociolinguistics; the previous avatar was of course one of the foundational
spirits in the broader Weld of sociolinguistics itself. As the title of his collected
papers of that vintage suggests (Gumperz 1971), that previous self was
(amongst other things) interested in how social groups express and maintain
their otherness in complex societies. Gumperz started as a dialectologist inter-
ested in tracking down the forces of standardization and particularly those of
diVerentiation, and it was the search for where these forces are located that has
led him inexorably from the macro-sociological to the micro-conversational
perspective; it was a long journey from the study of regional standards, to
ethnic groups, to social networks, to the activation of social boundaries in
verbal interaction, to discourse strategies. Readers will not understand his
work if they view it just as the study of conversation. He has not abandoned his
interests in the macro — it is the large-scale sociological eVects of multitudes
of small-scale interactions that still partially fuel his preoccupations with
conversation (see his contributions to Gumperz and Levinson 1996), most
evident perhaps in his concern with the plight of the individual caught up in
these large-scale forces. Those seeking to understand his attempt to link
macro and micro aspects of language use may still usefully turn to his studies
of code-switching, where many of his current ideas were adumbrated (see
especially Blom and Gumperz 1972). It was clear to him, for example, that
social situations partially determine the choice of code, and yet that code-
choice can partially determine situation — and this play between the presup-
posing and creative aspects of linguistic choice (to employ Silverstein’s termi-
nology) still dominates his work.
This background will also help to explain what may otherwise be taken as
a demerit: Gumperz’s analyses of conversations have nothing of the theoreti-
cal cleanliness to be found e.g. in conversational analysis. His tools are eclec-
tic, and the toolbox cluttered, on the one hand with pragmatic notions like
implicature, speech acts, frames, activities, cues, indices, and the like, and on
the other with sociological notions like network, ethnicity, gatekeepers, habi-
tus and so on. He is trying to depict processes that still defy understanding
with the best tools that come to hand from whatever school of analysis. And he
is trying to connect levels of analysis, from macro to micro, rather than to
develop an isolated level of conversational analysis. This is the real diVerence,
it seems to me, between his approach and e.g. conversation analysis, rather
perhaps than the one he oVers in the interview (in terms of his own preoccu-
pations with “situated on-line interpretation”; elsewhere he oVers a character-
ization of his own work in terms of the “social import of the Wne details of
verbal communication”, which seems more accurate).
With that as background, let me turn to the subject of Gumperz’s notion
of contextualization cues, which seems to me the central innovation in his
analysis of discourse. First, some remarks on the motivation. At the time he
was developing these ideas, I was lucky enough to be a graduate student in the
Language Behavior Research Lab at Berkeley where he worked. In Berkeley at
that time there was a rare and wonderful conXuence of ideas from diVerent
disciplines concerning the study of meaning — in philosophy, Grice and
Searle were expounding the ideas about implicature and speech acts now
associated with them, Fillmore was preoccupied with indexicality in language,
Kay with its sociological import, Robin LakoV with contextual meaning, and
George LakoV was attempting to wrap it all up in a uniWed theory of generative
semantics. It was an era of optimistic open-mindedness in which, for example,
Harvey Sacks could give an extended series of lectures to the Linguistics
Institute of the Linguistic Society of America. In Gumperz’s lab, we were taping
whatever we could get access to, and taping also in Weld locations around the
world, and we were trying to apply these theoretical ideas to the analysis of
actual snippets of conversation. It wasn’t easy. There was a yawning gulf
between what, on a simple-minded analysis, ‘the words meant’ and what we
took the participants to be self-evidently doing with their words. The gulf was
not to be bridged completely with the apparatus at hand — implicatures,

indirect speech acts, frames, and the like. This was the problem that preoccu-
pied Gumperz and his students at the time.1
In many ways, Gumperz’s approach to discourse analysis still has this
interpretative gulf as its primary theoretical target. He has since tried to
narrow the gap in a number of innovative ways, which are the hallmark of his
current approach. One line of attack was the careful analysis of prosody, the
neglected acoustic cues that might help to explain how we can possibly mean
so much by uttering so little. Another line was the apparently paradoxical idea
that utterances could somehow carry with them instructions about how to
build the contexts in which they should be interpreted. The two were com-
bined in the idea of a contextualization cue, often (but not necessarily) a
prosodic trigger that in conjunction with lexical material will invoke frames
and scenarios within which the current utterance is to be interpreted as an
interactional move.
I would like now to attempt to further clarify the notion of a contex-
tualization cue, which is perhaps better exempliWed rather than analytically
explicated in Gumperz’s work.2 There are two issues in particular worth
exploring. One is what the notion presupposes about the nature of context.
Another is the role of an implicit distinction between foreground and back-
ground information in messages. Out of these considerations should emerge a
clearer delimitation of the notion of a contextualization cue. For current
purposes I will assume a deWnition of context as a set of propositions taken for
granted by the participants3 (See especially Duranti and Goodwin 1992).
Let me turn then Wrst to the apparent paradox that utterances can create
their own contexts. The paradox would be: if it takes a context to map an
interpretation onto an utterance, how can we extract a context from an
utterance before interpreting it? The idea that utterances might carry with
them their own contexts like a snail carries its home along with it is indeed a
peculiar idea if one subscribes to a deWnition of context that excludes message
content, as for example in information theory. Context is then construed as
the antecedent set of assumptions against which a message is construed. But it
has long been noted in the study of pragmatics that this dichotomy between
message and context cannot be the right picture. We may want to say that use
of an expression of the form ‘The so-and-so’ presupposes the mutual knowl-
edge that there is one relevant so-and-so, but then Wnd that in fact a standard
way of informing an interlocutor that there is such a so-and-so is to say e.g.,
“The King of Tonga will begin his state visit on Monday.” Some have argued
that this is as it were an abuse of the conventions governing the use of deWnite
descriptions which forces the interlocutor to “accommodate” the utterance by
interpreting it as if it were mutual knowledge that there is a King of Tonga. But
in fact there are many other (non-presuppositional) devices that work in this
way — conventional implicatures associated with expressions like ‘but’ or
‘even’ force the creative construction of a context for the interpretation of the
utterances they occur in. “Even Harry will come” projects a context in which it
is assumed that there is a ranking of people in terms of their likelihood of
coming, and Harry is very low on that scale. Or suppose I say, “What are you
doing tonight?” — this projects some kind of invitation or request sequence in
the oYng.
So the idea that utterances can carry their contexts with them, that is, the
set of assumptions necessary to unpack their interpretation, is not as outland-
ish as it may seem at Wrst sight. The paradoxical quality of the idea of a
contextualization cue is as much due to our wanting to hang on to the simple
information theoretic view of what a context is. But still, there is a puzzle: how
does it work? Here both the phenomenon and the theorists part company.
Perhaps we can distinguish diVerent species of context-invoking aspects of
utterances:
1. Context-importation by conventional coding devices
Examples would be presupposition triggers like deWnite articles, expressions

which carry conventional implicatures like ‘even’ or honoriWcs in those lan-
guages that have them. Contrastive stress also Wts here: “It wasn’t me that did
it” projects a context in which someone is supposing I was indeed responsible.
Many languages have particles or other morphemes that serve the same sorts
of function: indicating that something is already presumed, or that no-one
assumes it, or that it is being introduced for discussion, or that its veracity is in
doubt, etc.. Interacting closely with the central meaning (e.g. truth-condi-
tional content) of the utterance, these conventional codings signal extra
propositions that form the background to the interpretation of the utterance.
2. Context-invocation by inference alone
Classical Gricean implicatures would be the prototype here:

(1) A: Hey, how about supper together?
B: I have a jealous husband
where in order to infer a (negative) response to the invitation, it is necessary to

invoke a large series of contextual assumptions of the sort: “One does not
choose to make one’s husband jealous”, “Eating supper is an intimate act” and
the like.
3. Context-invocation by “cue” or “Xag”
This is, I take it, the Gumperzian notion, in which the term “cue” denotes an
encoded or conventional reminder, like a knot in a handkerchief, where the
content of the memo is inferentially determined. Thus the “cue” cannot be
said to encode or directly invoke the interpretative background, it’s simply
a nudge to the inferential process. Moreover, the interpretative process is
guided more by a series of nudges now in one direction and now in another —
thus “cues” come as complex assemblages where the result of the whole
assemblage cannot be equated with the inferential results that each part alone
might have. The interpretive process may be guided by general pragmatic
principles of a Gricean sort, and thus be in many ways universal in character;
but the “cues” are anything but universal, indeed tending toward sub-cultural
diVerentiation. Hence the Gumperzian perspective on communication: at
once potentially possible across cultural divides and inevitably thwarted by
cultural nuances.
Further insights into how contextualization cues work, and their place in
an overall pragmatic scheme, may be found by turning to the second issue, the
relation between “foreground” and “background” in message structure. By
these terms I mean something entirely pretheoretical, the opposition between
central message content, coded propositional information, and peripheral,
more loosely associated and less clearly formulatable information, a sort of
informational penumbra. The opposition has aspects at diVerent levels: form,
content, and cognitive saliency (Table 1).
Table 1. The opposition between central and peripheral information.

Foreground Background
Form lexico-syntactic particles, modiWers,

prosody, kinesics
Content propositional general/vague or
non-propositional
communicative meta-communicative
Cognition salient inconspicuous
conscious unconscious
This set of alignments is the intuition in the Batesonian analysis of communi-

cation, with the diVerentiation of channels between conscious ‘verbal’ coding
and less conscious, less fully-coded prosodic and paralinguistic channels,
carrying with them some diVerentiation of function. Silverstein (1981) has
oVered some further elaboration of the relation between formal factors and
cognitive saliency: he suggests that certain factors tend to render aspects of
messages “out of awareness”. Thus non-segmentable, discontinuous, non-
iconic forms are likely to be inaccessible to native intuition; they will also tend
to be associated with non-referential, contextual content.
The message vs. context opposition is, as we have noted, a false opposi-
tion: the message can carry with it or project the context. But we tend to hang
on to the opposition because we focus on the foreground or message content,
and it is the background that tends to project the context. “Contextualization
cue” is one of a number of terms of art that attempt to explicate this relation
between message-background and context-projection. The hypothesis may
now, I think, be clariWed by suggesting that contextualization cues form a
natural class with the following cluster of properties:
Contextualization cues
Formal properties
1. a tendency towards non-segmentable, prosodic, paralinguistic, or kinesic
features;
2. if cued in lexico-syntax, then by lexical alternate (register) or minor gram-
matical class (e.g. particles);
3. any one clear function associated with a whole cluster of disparate features
(cf. Silverstein’s discontinuous feature), such features often being cross-
channel (a constellation of e.g. kinesic, prosodic and lexical features).
Content properties
1. “out of awareness” background features; they are context-invocative, and
cannot therefore be easily directly responded to;
2. non-propositional content, e.g. aVectual, rhetorical, social or metalin-
guistic;
3. tendency to invoke holistic bodies of assumptions (contextual “frames”),
which then play a role in the interpretation of the utterance — e.g. help
select reference, clarify rhetorical structure, indicate illocutionary force,
etc.;
4. content not really coded, but “cued” — i.e. reliant on large dose of inferen-
tial reconstruction; thus the inferred content of the same cues can be
diVerent in diVerent utterances.
The reasons why Gumperz has made this notion central to his analysis of
discourse should now I think be clear. It is because he is interested in the
relation between the micro- and the macro-sociolinguistic that contextua-
lization cues have a special interest. First, their “out of awareness” features,
coupled with their essentially arbitrary but loose association with formal cues,
mean they can only be learnt by rich exposure to a communicative tradition, a
deep immersion in social networks. This takes us back to the earlier Gumperz
with his interest in social groups and their networks. Second, in the workings
of contextualization cues can be found the springs of dialect and group diVer-
entiation: we pseudo-speciate by slow degrees that start here in subtle mis-
communications. In Labov’s view, in contrast, sociolinguistic diVerences are
little carriers of prestige or stigma. In Gumperz’s view, the smallest formal
diVerences may carry with them a chasm of incomprehension, because con-
textualization cues invoke the essential interpretive background for the fore-
grounded message. This is where the barriers Wrst come up, later to grow into
the saplings of dialects or the oaks of languages. This is also the last place for
the barriers to come down, as demonstrated in his analyses of the mini-
tragedies — the failed job application, the lost welfare beneWt — which have
been a favourite theme of his recent work, and which can be seen to be an
almost inevitable outcome of diVerentiated socialization. Gumperz is Ameri-
canized enough to put forward here an optimistic message about the educabil-
ity of the gatekeepers in multi-ethnic societies; others will Wnd here a deeply
pessimistic story about how elites guard access to social advancement.

A Wnal remark. In the interview in this volume, one of the interviewers
persists with a line of questioning that Gumperz seems to sidestep. If under-
standing is so complex, is it not ineVable? In particular, what gives the analyst
the right to say, “A intends this, but B thinks he means that”? Gumperz
suggests that it is all a matter of good ethnography: one asks A about what he
meant, and one notes that elsewhere C said something similar with similar
intent, and so on. But the interviewer, in good postmodernist style, is not so
easily assuaged: after all, if contextualization cues are often ambiguous, even
properly socialized ‘natives’ may understand diVerent things by the same
utterance. Gumperz assents, but does not draw the postmodernist conclusion:
he thinks that in certain cases, things are reasonably clear; he thinks he can
demonstrate a recurrent tendency by those who share the right kind of net-
work to associate a class of interpretations with a highly speciWc set of linguis-
tic cues invisible to those who belong to other networks. Radical doubt does
not assail him. In this, Gumperz is again a cheery optimist, and perhaps that
stems from his own Wrsthand experience as a network hopper, who landed on
American shores a refugee in his late teens.
In the twenty years since we sat together in the basement that served as the
Language Behavior Research Lab and pondered the gulf between the said and
the unsaid, Gumperz has surely but steadily tried to build out the piers of an
analytical bridge across it. He will be the Wrst to admit that the gap is still very
much there: we have today at best only the feeblest of understandings of
inferential processes in conversation. In a way, his own contributions only
deepen the puzzle about how human communication is possible, by drawing
attention to yet further layers of hidden, self-invoking machinery of the kind
sketched in his contextualization cues.
Notes
1. My own response, with Penelope Brown, was to develop a theory of politeness that we
hoped would help to bridge the gap between the said and the unsaid. Other students of his,
like Susan Gal and Deborah Tannen, have developed their own responses.
2. One of the clearest expositions may be found in Gumperz (1992).
3. There are all sorts of things wrong with this, but that is another issue.
References
Blom, Jan-Petter and Gumperz John J. 1972. “Social meaning in linguistic structure: Code-
switching in Norway”. In Directions in Sociolinguistics: The Ethnography of Communi-
cation, J.J. Gumperz and D. Hymes (eds.), 407–434. New York: Holt, Rinehart and
Winston.
Gumperz, John J. 1971. Language in Social Groups. A.S. Dil (ed.). Stanford, CA: Stanford
University Press.
Gumperz, John J. 1992. “Contextualization and understanding”. In Rethinking Context:
Language as an Interactive Phenomenon, A. Duranti and C. Goodwin (eds.), 229–252.
Gumperz, John J. and Levinson, Stephen C. (eds.). 1996. Rethinking Linguistic Relativity.
Silverstein, Michael. 1981. The Limits of Awareness. Sociolinguistic Working Paper 84.
Austin, Texas: Southwest Educational Development Laboratory.
Chapter 4
Contextualization and
social meaning-making practices
Paul J. Thibault
1. Preliminary considerations
My perspective is in broad sympathy with Gumperz’s attention to “the de-

tailed analysis of communicative practices.” However, my purpose in what
follows is both to suggest possible areas of theoretical convergence, as well as
to suggest alternative theoretical solutions where I see these as appropriate.
Gumperz draws our attention to the need for a theory of communicative
practice, or, in the terms that I use, “social meaning-making practices” (Lemke
1990; Thibault 1991). This emphasis on the communicative practices of a
community highlights the need for a theory of how linguistic and other
semiotic forms are used by the members of a given community, and why, in
the enacting of the social life of particular communities. To achieve this goal,
Gumperz is primarily concerned with the analysis of instances of social inter-
action, or “speciWc speech events” (Prevignano and di Luzio, this volume: 8).
He posits the need for some intermediate level of analysis which is able to
bridge the gap, both theoretical and methodological, which has all too often
seen the ‘linguistic’ and the ‘social’ banished to separate, seemingly incom-
mensurate domains of inquiry. In so doing, Gumperz shows that the descrip-
tion of a community’s communicative practices cannot adequately be
accomplished within the conWnes of any single discipline in the human and
social sciences. Such an enterprise is necessarily a transdisciplinary one, draw-
ing on the insights of sociology, ethnology, linguistics, anthropology, social
psychology, and so on, in order to develop a uniWed conceptual framework for
talking about social meaning-making (Gumperz 1992).
Gumperz’s focus on communicative practices is always conducted on
the basis of “independent, ethnographically-based analysis and information”
(Prevignano and di Luzio, this volume: 22–23). This underscores the “se-
mantic importance of context” (Prevignano and di Luzio, this volume: 9).
The notion of “communicative practice” refers, then, to the real-time pro-
cesses whereby participants construe semantic signiWcance in the unfolding
speech event. Gumperz emphasizes how the participants — speakers, listen-
ers, audience members — actively and jointly construct the speech event.
Meaning-making is fundamentally dialogic in character. He criticizes the
structuralist tradition of phonological and syntactical analysis in so far as
this has failed to provide the analytical tools and conceptual frameworks
necessary for the analysis of “everyday talk” (Prevignano and di Luzio, this
volume: 8–9). In pointing out that “Saussurean phonological and grammati-
cal structures, deWned in terms of Wnite sets of oppositions” (Prevignano and
di Luzio, this volume: 9), cannot by themselves provide an account of com-
municative practice, Gumperz pays somewhat less attention to the impor-
tance of accounting for the semiotic resource systems which members
deploy in their enactment of particular communicative events. Gumperz has
written of the ways in which “the localized choice of one or the other of the
paradigmatic set suggests diVerent implicatures in terms of which the utter-
ance at hand is to be interpreted” (1992:46; di Luzio, this volume). However,
there is no detailed speciWcation of these paradigmatic resource systems and
their relations to actual instances. I shall now discuss this question.
2. Social meaning-making: Resource systems and practices
How may the gap between the linguistic and the social in both linguistics
and social theory be overcome? In order to answer this, the relationship
between the semiotic resource systems of the community and its communi-
cative practices requires rethinking. This is especially so if we wish to avoid
being trapped by the abstract and de-contextualized formalisms in terms of
which language has been characterized as a closed and self-regulating system
of abstract forms.
If the detailed analysis of communicative practices is to “illuminate basic
issues in social theory” (Prevignano and di Luzio, this volume: 7) — that is, if
it is to show, for instance, how social ideologies are constructed and main-
tained in and through speciWc meaning-making practices, then we also need
an account of what it is possible to mean in a given community. In other
words, we need a complementary account of the social semiotic resource

systems that are available to the members of the community — the systems of
relations of semiotic forms of all kinds to their possible meanings. That is, how
do lexicogrammatical and other semiotic forms enable us to mean, and also,
what is it possible to mean in and through their use in discourse? From the
system perspective, the early attempts by Sapir and Whorf to construct an
account of language form and function which is informed by the insights of
cultural anthropology and Saussure’s theory of the system of value-producing
diVerences in langue, far from subscribing to context-free rules according to
the formalistic tendencies of early structuralism, already go in this direction
(Thibault 1997).
In his work on speech genres, Bakhtin emphasized that the study of
“grammar and lexicon” (1986:66–7), on the one hand, and the study of the
stylistics of the contextualized utterance, on the other, should not be opposed
to each other. This is so, Bakhtin points out, because “the speaker’s selection of
a particular grammatical form is a stylistic act” (ibid:66). The choice of par-
ticular options from the system of possible selections and their combinations
always constitutes a response to speciWc contextual contingencies. In ack-
nowledging this, Bakhtin further argues for the need to “organically combine”
the two perspectives, without, however, losing sight of the methodological
distinctions between them (ibid:66). Bakhtin’s solution to this problem lies in
his notion of speech genres. Speech genres specify the typical ways in which
the linguistic and other semiotic resource systems are deployed as regular,
repeatable, semiotic action formations in the social practices of some commu-
nity (Lemke 1988a; Thibault 1990).
These fundamental insights of Bakhtin’s have been further developed in
recent functional theories of lexicogrammar and discourse (Halliday 1994;
Martin 1992). Bakhtin does not himself provide any detailed account of these
resource systems. Further, the recent reception of Bakhtin in literary and
cultural theory has tended to emphasize heteroglossic diversity and stylistic
speciWcity at the expense of the typical and the systematic. Grammar and
lexicon, or lexicogrammar, belong to the system of language. The stylistics of
the individual utterance, to use the Bakhtinian terminology, refers to the
speciWc text or occasion of discourse. Actual speech events — cf. utterances —
are always speciWc semantic construals of the contexts which, as Gumperz
points out, they help to construct and deWne. As such, speech events respond
to speciWc semiotic and material contingencies, and in ways that are, at some
speciWable level of delicacy, unique to the particular context.

There is not, however, a direct, unmediated translation of the order of
system — lexicogrammar — into speciWc occasions of discourse. Rather,
genres are types. But they are types in a rather particular way. Genres do not
specify the lexicogrammatical resources of word, phrase, clause, and so on.
Instead, they specify the typical ways in which these are combined and de-
ployed so as to enact the typical semiotic action formations of a given commu-
nity. The ‘translation’ from lexicogrammatical system to speciWc speech event
is always mediated by the typical patterns of language use — the primary and
later the secondary genres — in the community. The notion of speech genre is
an intermediate level of analysis in this sense.
How are the resources of the system integrated in the joint enactment of
some semiotic performance, or speech event? My starting point here is that
all communicative events are multimodal semiotic performances (Lemke
1998; Thibault 1994; Van Leeuwen 1992). That is, they co-deploy in
an integrated way the resources of diverse semiotic modalities — lexico-
grammar, prosody, posture, kinesics, rhythm, and so on. The meaning of the
overall discourse event cannot be localized in any single modality, seen as
isolated from the others. Gumperz rightly rejects the notion of discourse as a
mere string of clauses. Taken literally, such a view would amount to a linear
and additive view of meaning-making. However, meaning-making does not
occur on this basis. Rather, the resources of diverse semiotic modalities mu-
tually contextualize each other in a multiplicative way (Lemke 1998). Lan-
guage and other semiotic modalities are not, for this reason, independent of
each other to start with. This is so both synchronically and diachronically:
language both co-evolved with the other modalities of meaning-making and
is co-deployed with these. A given discourse genre consists of typical co-
patternings of selections from a diversity of semiotic resource systems. In
drawing attention to “communicatively signiWcant prosodic and paralinguis-
tic aspects of speech” (Prevignano and di Luzio, this volume: 22), Gumperz
points to the ways in which selections in one semiotic modality may aVect
the meaning of a selection in some other, or how the new contextual redun-
dancy which results from the co-patterning of the two may result in a newly
contingent contextual meaning. That is, one which is not predicted by the
usual patterns of the two selections, taken ‘separately’.
3. Context-dependent and context-independent meanings
Gumperz’s solution to these issues lies in what he calls a “move from a

Saussurean to a more inclusive, broader, Peircian semiotics” (Prevignano and
di Luzio, this volume: 10). In my view, the distinction between symbolic and
indexical signs is itself problematic. Gumperz deWnes the notion of “con-
textualization cue” as a class of indexical sign. He contrasts the ‘pure’ indexi-
cality or context-dependent character of these to the purportedly context-free
lexical and propositional meanings of the symbolic sort.
The distinction between symbolic and indexical signs thus rests on the
assumption that there are two very general classes of meanings, viz. those of
the context-independent sort and those which are context-dependent.
Broadly speaking, the former rest on the assumption that there is such a thing
as a context-free or ‘literal’ propositional meaning. There is an assumption of
a Wxed, literal correspondence relation between the propositional content of
the form and some extralinguistic state of aVairs to which the proposition
refers. Indexicals, on the other hand, are seen as being, to varying degrees,
non-propositional and context-dependent.
But the distinction between symbolic and indexical signs is itself problem-
atic in the following two ways. First, the notion of “context-free lexical mean-
ings”, based on “well-known grammatical and lexical rules”, derives, in actual
fact, from a speciWcally western linguistic folk-theory of reference (Silverstein
1987; Rumsey 1990) and the various philosophical reWnements of this.1 Ac-
cording to this folk-theory, some classes of surface segmentable lexicogram-
matical forms — e.g. nouns — when abstracted from context, may be assigned
a Wxed word-object correspondence relation. This is a system view of sorts in
the sense that the word-object correspondence relation is seen to apply to
abstract noun classes, abstracted from their contexts-of-use. Furthermore, it is
a view which recognizes just one kind of structuring principle in language, viz.
constituency, along with the bias towards typological-categorial meanings
that this entails. In this view, the meanings of linguistic forms can be glossed
according to context-free principles of referential equivalence between lan-
guage form and extralinguistic states of aVairs. Language forms so deWned are
explicitly meaningful for the insider-members of the culture, as witnessed by
the way in which such forms may be assigned a dictionary meaning as lists of
formal class items, irrespective of context-of-use.
Now, meanings of the so-called context-dependent or indexical sort tend
not to be so readily segmentable into discrete particles or constituents of

isolable linguistic form such as words, phrases, clauses, and so on. They tend
to be “analogic” and continuous, rather than “digital” and segmental, in
character (Bateson 1973:261–2). Thus, they conform to quite diVerent orga-
nizational principles. These tend to be topological and continuous in char-
acter, rather than typological and discrete.2 From the perspective of
insider-members, the resources of these semiotic modalities are ‘felt’ or ‘intu-
ited’ as having implicit meaning in some speech event, even though members
Wnd it diYcult to recognize or to gloss such meanings as being explicitly
communicative. SpeciWcally, they cannot usually be analyzed as explicit
chunks of digitalized experiential content. However, and as Gumperz’s work
shows, prosodies and paralanguage are always communicative from the point
of view of the analyst, who is committed to reconstructing the underlying
principles of contextualization in discourse.
The distinction between symbolic and indexical signs serves in my view to
highlight the need to give equal analytical and theoretical value to both the
typological-categorial distinctions in terms of which language, seen as au-
tonomous with respect to other semiotic modalities, has usually been deWned,
and the topological-continuous character of not only many other, frequently
neglected dimensions of linguistic organization, but also of the other semiotic
modalities with which language is always integrated in social meaning-mak-
ing. Thus far, our theories of both language and other semiotic modalities
have been overwhelmingly shaped in terms of the former. This has meant that
linguistic structure has been seen as prototypically typological-categorial and
digital in character, as witnessed by the almost exclusive reliance on part-
whole constituent hierarchies in the analysis of linguistic form. At best, this is
a singularly partial view of the object of study.3
The Wrst problem resides, then, in the implicit assumption that “context-
free lexical items” may be described from the perspective of the system of
linguistic forms, abstracted from their speciWc contexts-of-use, whereas other
semiotic modalities may not be so described because these are merely context-
dependent indexes of the given situation. In this view, the latter would have no
internal systematicity of their own. I doubt very much that this is Gumperz’s
own position, though his focus on instances may be seen to invite this inter-
pretation. From the system point of view, the meaning-making resources of all
semiotic systems are analytically reconstructable as abstract systems of diVer-
ences which cross-classify with each other in potentially very many diVerent
ways (see above). There is no reason in principle why, for example, the
gestural semiotic modality, intonation, or depiction cannot be described as
systems of diVerences, or as systems of possible form-meaning relations, seen
as the resources for making context-dependent meanings in these modalities.
These may have proved intractable to the analytical methods and procedures
of traditional (structuralist) linguistic analysis. However, there is no reason
why a system of gestural diVerences, say, cannot be postulated so as to specify
the possibilities for meaning-making in that semiotic modality.
In any case, the assumption that lexical items can be speciWed on the basis
of Wxed form-meaning relations is itself questionable. From the point of view
of the system, semiotic forms of all kinds — linguistic, gestural, and so on —
have a meaning potential (Halliday 1978). In fully contextualized discourse,
there is a great deal of semantic variation in the use of the ‘same’ lexical item
according to factors which may relate to genre, for example, or to the semantic
relations that may be made between items in speciWc contexts. The system of
diVerences, seen as abstracted from speciWc contexts, is variably deployed
according to speciWc contextual requirements.
The second problem concerns the folk-linguistic view in our (western
European) cultures and their historical oVshoots that linguistic forms are only
explicitly communicative for insiders if they can be glossed as having a Wxed,
de-contextualized propositional meaning which functions to “refer to” desig-
nated states of aVairs in the real-world (Silverstein 1987).4
Now, from the point of view of the outsider-analyst of social meaning-
making, all semiotic forms are both meaningful and communicative, irrespec-
tive of whether they are explicitly communicative for insider-participants or
not. Once we drop our folk-theoretical Wction that some classes of lexicogram-
matical forms, prototypically nouns, may be assigned a Wxed referential mean-
ing, irrespective of context, then it should become clearer that, in the context
of a speciWc speech event, it makes no sense to uphold a distinction between
context-independent symbolic signs and context-dependent indexical ones. If
all semiotic modalities may, from the system point of view, be speciWed as
systems of diVerences, abstracted from speciWc contexts, then it is no less
important to show how it is through the co-deployment of these resources
that participants jointly enact the speech event. It is generally accepted that
intonation, say, can be analyzed as a system of potential meaning-making
diVerences which systematically contrast with each other, viz. rising, falling,
falling-rising, rising-falling tone, and so on. Abstracted from speciWc dis-
course contexts, these can only have a meaning potential rather than a fully
contextualized discourse meaning.5
On the other hand, all semiotic forms are indexical when deployed in
some context. Further, the resources for indexing a particular form in a
contextually speciWc way may themselves be speciWed systemically. For ex-
ample, this is so of the often very complex systems of deictic categories in all
languages. In English, for example, the systemically speciWable diVerences
between ‘the’, ‘a’, ‘this’, ‘that’, and so on, specify how the Head noun in the
nominal group is grounded in relation to some material or other feature of the
situation which is made semiotically salient in the interaction. If I say, for
instance, “Look at that man over there”, to index a particular man in the
spatio-temporal purview of the speaker and listener, then I am using a selec-
tion from the system of deictic categories in English — ‘that’ — to index the
perceived object as an instantiation of the type-category speciWed by the Head
noun ‘man’ in the nominal group in question. Furthermore, I am doing so in
ways which index the man in question as a semiotically salient feature of the
interaction. The man is made a relevant part of the discourse context. From
the system perspective, the noun ‘man’ construes a type-category which is
internal to the grammar and semantics of the English language. As such, it
does not refer to any speciWc ‘real-world’ man. It is only when it is grounded
by the resources of deixis that the noun speciWes particular states of aVairs,
objects, events, and so on as instantiations of the type-category. Further, there
may be degrees of conformity to the criterial features speciWed by the type-
category (Langacker 1987:68–71).
A noun, as a grammatical class item, speciWes a type-category of a seman-
tic ‘thing’ which belongs to a given language system. Its indexical grounding
with respect to a given material or imagined object, event, etc. means that the
perceived or Wctive phenomenon so indexed through the resources of the
nominal group is construed as an instantiation of the type-category. There is
no direct word-object relation to start with. Nouns do not simply refer to
objects, states of aVairs, etc. in a direct, unmediated way. Rather, they specify
an experiential category which is internal to the language system. The
diVerence between signs with experiential content (e.g. nouns) and those with
none (e.g intonation) is one of degree. It is not an absolute distinction. In
discourse, all signs function in one way or another to index the wider situ-
ational context. Both lexicogrammar and intonation can be analytically re-
constructed as systems of contrasting form-meaning diVerences. In discourse,
what is important are the cross-coupling patterns between (i) the diverse
semiotic modalities in operation, and (ii) these and the material world in the
enactment of the wider situational context (Lemke 1993:250; Thibault
1996:342). Informally, some examples are:
1. (nominal group / perceived object or event) // situation)
2. (lexicogrammar / intonation) // aVective-attitudinal orientation of speaker)
3. (phoneme / morpheme category) // semantic concept)
These examples illustrate, albeit informally, how local cross-coupling patterns
between, for example, a given nominal group and some perceived phenom-
enon in the material world, or between some choice in lexicogrammar and
intonation, combine to index some aspect of the wider context which the
combining of the two modalities helps to constitute. In each case, there is a
Wrst-order contextual redundancy pattern between the selections from the
two modalities. This is represented by the single slash, viz. ‘/’. Yet, in each case
it is necessary to talk about a second-order relation which speciWes the wider
context in which this particular combination has the meaning it does. This is
represented by the double slash.
But notice that the third of these three examples refers to the language
system in the narrow sense. The same basic principle applies here too. Thus,
the co-patterning of a particular phoneme category with a morpheme is a
Wrst-order contextual relation on the basis of which a second-order semantic
concept is construed in the language in question.
In each case we see an important principle at work. This is the principle of
‘what goes with what else’ in the enacting of the wider situational context in
which a particular combination has its meaning. An index was deWned by
Peirce as a sign which is “physically connected” to some object (Nöth
1990:123–4). If indexicality is centrally concerned with the making and speci-
fying of contextual relations, as Gumperz’s work shows, then this deWnition is
no longer very satisfactory for a theory of discourse. The general problem of
“indexicality” has to do with how particular cross-couplings among diVerent
semiotic modalities and between these and selected aspects of the phenomenal
world instantiate a local discourse context by specifying in various ways the
parameters of the interaction itself.
4. Discourse, context, and meaning-making practices: Towards an

integrated account
Gumperz draws attention to the shifting nature of participant interpretation

in the ongoing speech event. In doing so, he identiWes a number of ways in
which participants interpret and understand, in contextually constrained
ways, the nature of the discursive activities they jointly enact and engage in
with others. For my present purposes, I shall single out the following three
categories, as proposed by Gumperz (Prevignano and di Luzio, this volume):
1. contextualization cues, seen as propositionless pure indexicals;
2. the inferencing of extra-communicative background information;
3. genre as an ideologically-charged metapragmatic assessment for referring
to or labeling texts or speech exchanges.6
Each of these categories plays its role in explaining how the participants in
some communicative event interpret and understand the unfolding interac-
tion. Gumperz’s eclectic approach highlights two critically important issues.
First, no single approach or tradition in the human and social sciences can
on its own provide all of the analytical tools and theoretical categories that
are required in order to answer the kinds of questions that Gumperz raises
about discourse processes and social practices. Secondly, the attempt to inte-
grate the insights from various approaches and traditions means that an act
of reconstitution of their theoretical categories and analytical procedures is
called for. This is necessary in order to produce a conceptually uniWed
framework — one which can overcome the ideologically disabling disjunc-
tion between the “macro” and the “micro” approaches to the analysis of
social phenomena (Thibault 1991:229–40). In the process of borrowing and
adapting insights from various frameworks, these cannot simply remain as
they were. Rather, they must themselves be changed in the process of their
theoretical reconstitution.
In my view, the above three categories refer, in actual fact, to three essen-
tial parameters in terms of which all acts of social meaning-making may be
deWned. This is so irrespective of the speciWc semiotic modalities that are
deployed. On this basis, I now propose three generalized meaning-making, or
context-building, strategies, as follows:
I. Indexical meaning-making practices;
II. Intertextual meaning-making practices;
III. Metadiscursive meaning-making practices.
The three categories are analytically, though not constitutively, separable

dimensions of all acts of social meaning-making. The fundamental question at
stake here is: how do discourse participants integrate the diverse semiotic
resource systems that are deployed into a single semiotic performance? Or,
what are the contextualizing parameters in terms of which such integration is
achieved? In order to provide a coherent theoretical response to these ques-
tions, we need to specify a number of very general meaning-making strategies
that apply to all modalities of semiosis. I shall argue that the three parameters
proposed above satisfy this requirement. Further, they do so in a conceptually
uniWed way, as I shall now show.
I. Indexical meaning-making practices
Indexicality, or “grounding” (Langacker 1987:126–8) refers to the fact that all

semiotic forms and their functions in discourse contribute in some way to the
ongoing enactment and maintenance of the wider situational context of which
these forms are a constitutive part. From the instantial perspective of dis-
course, there is not, then, some special class of semiotic forms whose function
it is to indicate the context. Instead, all the forms used in conjunction with
each other function in various ways to enact and to specify the overall dis-
course event. From the point of view of discourse, the distinction between
context-independent symbolic signs and context-dependent indexical ones is
meaningless. All of the semiotic forms which are co-deployed in some com-
municative event serve a plurality of semiotic functions in that event. In so
doing, they act on and aVect each other in complex and multiple ways.
In my view, the distinction between propositional and propositionless
signs is better handled by the distinction that Lemke (1984:69) makes between
content-speciWc and type-speciWc semiotic acts (see also Thibault and Van
Leeuwen 1996:567–8). A content-speciWc act in some way categorially con-
strues or makes salient some speciWc features of the situation as being an
instance of an event-type, act-type, etc. that is categorized by the ideational
resources of the linguistic system and is, for this reason, recognizable in that
community. If I say, for example, “The cat is drinking the milk,” with refer-
ence to a particular material happening, then I am not only indexing a speciWc
cat in the situation through the combined use of speciWc deixis (‘the’) and the
type-category speciWed by the noun ‘cat’. I am also construing, in my use of
this clause, the perceived happening as an instantiation of a particular activity-
schema, as modeled by the semantic conWguration [Actor^Process: Material

Action^Goal] in the grammatical-semantic resources of the English clause. In
so doing, I am making some perceived phenomenon in the purview of speaker
and listener semiotically salient to the discourse event in a content-speciWc
way. In indexing this phenomenon as an instantiation of a particular activity-
type, I am constructing a contextually speciWc relation between the perceived
event and the particular clause selection that I use. The clause-level semantics
categorize or construe the material event as being the instantiation of a certain
type-category of event which is modeled in the semantics of English clause
grammar (Halliday 1994: chap. 5).
A type-speciWc act, on the other hand, is an act which is in some way
appropriate to the situation, irrespective of whether its ideational content, if it
has any, indexes an actual state of aVairs or not (Lemke 1984:69). Consider
the following, discussed in Levinson (this volume: 35):
(1) A: Hey, how about supper together?

B: I have a jealous husband.
B’s response is content-speciWc if it is interpreted as referring to an actually

jealous husband, at least from the point of view of B in the exchange. In this
case, B provides a Reason-Explanation for declining A’s invitation. On the
other hand, B’s response is type-speciWc if, for example, she uses it as a
conventional strategy for forestalling the strategically calculated plan which
she may interpret as motivating A’s invitation. In this case, B’s speech act is a
conventional strategy which may be used to decline such invitations, and
without giving oVence to the proposer, irrespective of whether her husband is
actually jealous or not. B’s response to A may be interpreted as being either
content-speciWc or type-speciWc or, perhaps, both.
Whatever interpretation prevails is always a question of the ways in which
the contextually constrained meaning potential of an utterance is taken up
and negotiated in the unfolding discourse (Thibault and Van Leeuwen
1996:577–82). Semiotic acts, whether content- or type-speciWc, always con-
tribute to the joint enactment of the relevant context. The particular examples
I have focussed on may appear to exemplify the symbolic or propositional
type of meaning discussed above by virtue of their exhibiting clause-level
propositional content. In actual fact, these examples do not easily conform to
the notion of context-free lexical and grammatical rules. I have tried to show
that these, too, always function to index some aspect of the relevant contextual
parameters whereby a particular communicative event is enacted.
II. Intertextual meaning-making practices
Gumperz refers to the extra-communicative “background information”

which discourse participants draw on in order to make appropriate interpre-
tative inferences. This raises two orders of problems. First, the notion of
background information is not, analytically speaking, close enough to the
order of discourse to be really useful. In my view, Gumperz is talking about a
most important dimension of all social meaning-making, and one which is
not speciWable at the level of, say, lexicogrammar per se in the linguistic
system. Secondly, there is a danger that the notion of background information
is reiWed as an abstract psychological property of the individual. In other
words, an essentially social order of meaning relations is seen as simply
already existing ‘in’ the minds of individual members of the culture in ques-
tion. On both counts, the speciWc examples which Gumperz discusses belie
these assumptions. With this in mind, I shall now propose an alternative to the
notion of background information — one which puts social meaning-making
in the center of the picture.
The central question to tackle is: how do discourse participants construe
meaningful relations between one occasion of discourse and another? Or, how
does one occasion of discourse serve as the context for the interpretation of
another? Following the work of Jay Lemke (1985), I shall discuss these ques-
tions in terms of the intertextual meaning-making practices (IMMP’s) of some
community (see also Thibault 1986, 1990). Two of the examples that Gumperz
himself discusses — viz. the announcer on the classical music radio station
and the rape trial transcript — provide a good starting point. Gumperz points
out how, in both cases, he had to “search [his] memory of past communicative
experience to construct a likely scenario or narrative plot that might suggest
possible interpretations” (Prevignano and di Luzio, this volume: 13). In both
cases, Gumperz’s own observations show how the speciWc meanings he as-
signs to these instances are construed in and through their relations to still
wider, more abstract, typical patterns of meaning relations which the mem-
bers of a particular community may share, access and draw on to varying
degrees and in varying ways.
Thus, the contrastive accent on ‘you’ in the radio announcer’s speech is
assigned its speciWc meaning in relation to expectations concerning the typical

patterns of relations that are construed between intonation and lexico-
grammar, and context-type. In this example, Gumperz draws our attention to
local features of the interactional context. It is the lack of expectational ‘Wt’
between the two semiotic modalities which provides the basis for the speciWc
meaning that Gumperz assigns to the event. This example shows how often
very subtle co-patternings of selections across semiotic modalities may be
crucial to the meaning of the discourse event, or to some local part of this.
In his second example — the rape trial transcript — Gumperz starts from
assumptions concerning the type of social activity which is taking place. In
contrast to the Wrst example, the perspective he adopts on this occasion is a
global one. SpeciWcally, he interprets the speech event through the genre of
“cross-examination as adversarial proceeding.” Again, Gumperz draws our
attention to the ways in which discourse participants relate the speciWc mean-
ing selections that are made — viz. the speech act strategies of attorney and
victim — to their expectations about the particular discourse genre through
which they interpret the activity which is taking place. It does not follow, of
course, that there is always consensus about this, as this example shows.
Both of the perspectives — local and global — illustrated in these two
examples draw attention to a second most fundamental dimension of social
meaning-making, viz. the IMMP’s of some community. Intertextuality does
not simply refer, as in much of literary theory and criticism, to the relations
between a particular text and some other, usually historical prior text (Thi-
bault 1991:135). The fact that it may do so is not at issue here. Literary
citation, allusion, paraphrase, and so on, are all intertextual in this more
restricted sense. More fundamentally, it concerns the multiple ways in which a
community’s meaning-making practices enable discourse participants to link
two or more texts or occasions of discourse on the basis of some more abstract
pattern of meaning relations that they are seen as having in common. Inter-
textual meaning relations are, then, abstract patterns of relations that dis-
course participants draw on both to construct and to interpret speciWc
instances. Thus, no text or occasion of discourse is ever understood in iso-
lation from the wider systems of intertextual meaning relations in the com-
munity (Lemke 1985:275). Rather, all discourse occasions are made and
interpreted in and through wider systems of IMMP’s. This does not mean that
these are shared, accessed, or interpreted in the same way by all members of
the community. Bakhtin’s (1981) notion of “social heteroglossia” represents
an important early attempt to explore the diversity of IMMP’s, and the rela-
tions of alliance, conXict, co-optation, and so on among these in particular
texts and occasions of discourse (Lemke 1988b; Thibault 1989).
As Gumperz’s examples show, IMMP’s entail more abstract, or higher-
order, relations of relations of relations ... in Bateson’s sense. They are neither
systems of ready-made text-types nor recipes showing how to construct par-
ticular types of texts. Rather, they provide an interpretative resource for
construing culturally salient patterns of meaning in speciWc instances. In
neither case is it a simple question of whether the speciWc instance matches the
intertextual pattern or not. Rather, discourse speciWc meanings are inter-
preted on the basis of whether and to what extent they conform to or contrast
with a given intertextual pattern. IMMP’s are neither homogeneous nor con-
sistent across the diverse practices that constitute a given community. That is
why they are said to be heteroglossic in Bakhtin’s sense.
The notions of “background information” and “inferential procedure”
risk both de-semioticizing what are in fact the systems of IMMP’s in a com-
munity and also reifying their essentially social nature as if these were proper-
ties of individual mind. Neither of these notions adequately speciWes the local
(instantial) resources whereby discourse participants reconstruct the relevant
IMMP’s as they deploy these in their own meaning-making practices. The
notion of IMMP’s better enables us to reconstruct in our analyses the diverse
ways in which members indicate which intertextual relations are relevant, and
how these relate to which other occasions of discourse (Lemke 1985:276).
Neither “background information” nor “inferential procedure” is ade-
quately grounded in a theory of the semiotic forms and their modes of deploy-
ment in discourse to bring about the conceptually uniWed account — one
which can reconstruct the links between resource systems and social mean-
ing-making practices — to make this possible. The idea of a ‘background’
certainly has the merit of suggesting the diVuse, taken-for-granted, and out-
of-focus assumptions that characterize the lifeworld or the habitus of values,
practices and orientations that inform the speciWc discursive event (see also
Levinson, this volume). As it stands, however, I remain unconvinced that it is
able to provide us with the means for analytically reconstructing the links
between the “macro” and the “micro” perspectives in a conceptually uniWed
way (see Thibault 1991:229–40 for further discussion).
III. Metadiscursive meaning-making practices
Gumperz also refers to Bakhtin’s notion of speech genres. This is the meta-
discursive dimension of social meaning-making. For Gumperz, genre is “an
ideologically charged metapragmatic assessment”, to echo a Silversteinian
turn of phrase, or “forms of talk about talk”, in the GoVmanesque parlance.
From this point of view, discourse participants make interpretative as-
sumptions about the meaning of their own and others’ contributions to the
ongoing discourse event. They self-reXexively interpret to varying degrees of
explicitness and awareness the meaning of particular discourse events or parts
of these, relative to some social viewpoint. To do so, members deploy the
meaning-making resources of the system so as to assign signiWcance to their
own and others’ local instantiations of these same resources. Social meaning-
making practices may be said to be metadiscursive in this sense.
Whereas IMMP’s connect the patterns in the speciWc instance to still
wider patterns of meaning in a given culture, metadiscursive meaning-mak-
ing practices (MMMP’s) serve to articulate the links between the local, or
occasion-speciWc, meaning selections in some speciWc communicative event
to the paradigmatically-organized contrast sets of context-types in relation to
which the former have their meaning. That is, MMMP’s link the speciWc
instance or meaning choice to the metasystem of possible alternative choices
from which it was selected. This may be so to varying degrees of completeness
and explicitness. MMMP’s bring into focus the meaning of the particular act
by providing a connection between this and the higher-order metasystem of
possibilities which gave rise to it.
However, it is not enough simply to say that metapragmatic assessments
involve “talk about talk”. There are many diVerent ways in which members
talk about talk. As analysts, we need to be able to specify which forms of talk
about talk occur, and in which context-types. What are the speciWcally ‘meta’
relations which specify the relations between the particular act, the context in
which it occurs, the always social viewpoints of the participants in the dis-
course, and the metasystem — the contrast sets — which any occasion of
metadiscourse always invokes? Whenever we engage in talk about talk, we
always invoke or reconstruct, to varying degrees of completeness, the meta-
system in terms of which the given act has its meaning. MMMP’s make
explicit in some way the fact that the metasystem is always immanent in
speciWc occasions of discourse. Thus, MMMP’s interpret or gloss the meaning
of the instance from the global standpoint of the metasystem. In making the
link between local occasion and system, participants show how they have their
own socially-constrained metatheories — folk or otherwise — concerning the
ways in which occasion-speciWc acts of meaning are relatable to the global
system of possibilities which engendered them.
5. Conclusion: Towards a functional semantics of action
The above three-way distinction refers to diVerent aspects of the constant

dialectic between local (instantial) and global (systemic) relations in dis-
course. These may be summarized as follows:
I. Indexicality is concerned with the ways in which selections from the global
resource systems of the various semiotic modalities cross-classify and inte-
grate both with each other and with selected features of the material-phenom-
enal world so as to enact a local, occasion-speciWc, discourse event;
II. Intertextuality speciWes the global systems of intertextual meaning relations

that are seen as relevant to the interpretation of a given local occasion of
discourse;
III. Metadiscursivity is concerned with the ways in which the global system of
possible sets of alternatives — the metasystem — is made locally salient in
some interaction. It does so by providing the means by which participants
construct a self-reXexive connection between the metasystem and some dis-
course event. This means that discourse participants draw on this system in
the process of assigning meaning to some discourse event. They do so by
explicitly focussing the metasystem onto the immediate interpretative prob-
lem. In this way, the relevance of the global possibilities of the metasystem is
made an explicit object of metasemiotic consciousness, relative to some par-
ticipant viewpoint.
The notion of contextualization is a fundamental one in Gumperz’s theory of

discourse. Gumperz has shown us how the ongoing interactional event is
constructed and construed in and through the ways in which the localized
choices — utterances, speech acts, and so on — of the interactants relate to
each other in the real-time unfolding of the discourse. This depends on the
ways in which interactants integrate all of the diverse contextualization cues
into their sense of the overall activity which is taking place (Gumperz
1992:44). Rather than trying to interpret the other semiotic modalities from
the exclusive standpoint of linguistic analysis, the task should be to develop
ways of grounding the multimodal nature of such events in a functional
semantic approach to action as a whole. The three-way distinction I made
above attempts to provide a suYciently generalized and integrated framework
for analyzing the local and the global dimensions of social meaning-making.
Gumperz’s work on contextualization makes a seminal contribution to the
development of a semiotics of action in which the functional integration of all
of the semiotic modalities which are involved is accounted for.
Notes
1. Thus, Wittgenstein in his Tractatus: “3.203 A name means an object. The object is its
meaning” (1969 [1921]:23).
2. See Kenneth Pike’s (1967) wave and Weld, as distinct from his particle, perspectives on
language, prosodies of all kinds, suprasegmentals, and so on.
3. It may well turn out, as further research is conducted on this question, that what we
conventionally refer to as ‘language’ and treat as a uniWed semiotic system will turn out to
be a diverse set of possibilities whereby linguistic forms are integrated with other semiotic
modalities. In the conventional view, speech and writing have been seen as diVerent
material realizations of the same lexicogrammatical forms in the oral-aural and optical-
visual channels. In my view, speech and writing are themselves folk-linguistic glosses on a
diverse set of multimodal meaning-making practices in which diVerent possible integra-
tions of lexicogrammatical, prosodic, paralinguistic, graphological, pictorial, kinesic, spa-
tial, rhythmic, and other resources characterize the discourse genres — the semiotic
action formations — of a given community. In such a view, genre will, increasingly, be
deWned as the typical modes of deployment of the multimodal meaning-making resources
available to the members of a given community.
4. Alan Rumsey (1990) points out how, in the Ungarinyin language of Australia, the
meanings of linguistic locutions are glossed, not on the basis of de-contextualized criteria
of reference, but, instead, on ‘pragmatic’ criteria concerning the ways in which the
locution, or some alternative, locally construed as equivalent to it, is used in the various
context-types of the culture.
5. Bolinger, for instance, recognized long ago that entire intonational conWgurations may
be related to speciWc context-types in culturally speciWc ways (1951:209). Further, and just
like lexicogrammatical forms, these are “ineVable” in the sense that they cannot be reliably
glossed in terms of some extra-semiotic absolute standard (Halliday 1988). Rather, the
meaning of a particular intonation selection itself derives in the Wrst instance from the
place which any given selection is felt to occupy in an overall system of contrasting options
which deWnes its semiotic value.
6. In each case, Gumperz acknowledges the diverse sources which have inXuenced his
thinking on these issues. In relation to (1) above, he singles out Peirce’s notion of indexical
signs and GoVman’s frames; he relates (2) to GarWnkel’s inferential processes; and (3)
derives in part from Bakhtin’s theory of speech genres, as well as Silverstein’s (1992:60)
“metapragmatics”.
References
Bakhtin, Mikhail. 1981. “Discourse in the novel”. In The Dialogic Imagination: Four Essays
by M.M. Bakhtin, C. Emerson and M. Holquist (trans.), M. Holquist (ed.), 259–422.
Austin and London: University of Texas Press.
Bakhtin, Mikhail. 1986. “The problem of speech genres”. In Speech Genres and Other Late
Essays, V.W. McGee (trans.), C. Emerson and M. Holquist (eds.), 60–102. Austin:
University of Texas Press.
Bateson, Gregory. 1973. “The logical categories of learning and communication”. In Steps
to an Ecology of Mind, 250–279. London and New York: Paladin.
Bolinger, Dwight L. 1951. “Intonation: Levels versus conWgurations”. Word 7 (3):199–
210.
Di Luzio, Aldo. “Presenting John J. Gumperz”. This volume.
guage, P. Auer and A. di Luzio (eds.), 39–53. Amsterdam and Philadelphia: John
Benjamins.
Halliday, M.A.K. 1978. Language as Social Semiotic. The Social Interpretation of Language
and Meaning. London: Arnold.
Halliday, M.A.K. 1988. “On the ineVability of grammatical categories”. In Linguistics in a
Systemic Perspective, J.D. Benson, M.J. Cummings, W.S. Greaves (eds.), 27–51.
Amsterdam and Philadelphia: John Benjamins.
Halliday, M.A.K. 1994. An Introduction to Functional Grammar. Second edition. London
and Melbourne: Arnold.
Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol.1. Theoretical Pre-
requisites. Stanford, CA: Stanford University Press.
Lemke, Jay. 1984. “Action, context, and meaning”. In Semiotics and Education. Victoria
University, Toronto: Monographs, Working Papers and Prepublications of the
Toronto Semiotic Circle 2:63–93.
Lemke, Jay. 1985. “Ideology, intertextuality, and the notion of register”. In Systemic
Perspectives on Discourse, Vol. 1. Selected Papers from the 9th International Systemic
Workshop, J.D. Benson and W.S. Greaves (eds.), 275–294. Norwood, NJ: Ablex.
Lemke, Jay. 1988a. “Genres, semantics, and classroom education”. Linguistics and Educa-
tion 1 (1):81–99.
Lemke, Jay. 1988b. “Discourses in conXict: Heteroglossia and text semantics”. In Systemic
Functional Approaches to Discourse. Selected Papers from the 12th International
Systemic Workshop, J.D. Benson and W.S. Greaves (eds.), 29–50. Norwood, NJ:
Ablex.
Lemke, Jay. 1990. Talking Science. Language, Learning, and Values. Norwood, NJ: Ablex.
Lemke, Jay. 1993. “Discourse, dynamics, and social change”. In Language as Cultural
Dynamic. Cultural Dynamics VI (1–2):243–275.
Lemke, Jay. 1998. “Multiplying meaning: Visual and verbal semiotics in scientiWc texts”.
In Reading Science: Critical and Functional Perspectives on Discourses of Science, J.R.
Martin and R. Veel (eds.), 87–113. London and New York: Routledge.
Levinson, Stephen C. 1997. “Contextualizing ‘contextualization cues’”. This volume.
Martin, James R. 1992. English Text. System and Structure. Amsterdam and Philadelphia:
John Benjamins.
Nöth, Winfried. 1990. Handbook of Semiotics. Bloomington and Indianapolis: Indiana
University Press.
Pike, Kenneth L. 1967. Language in Relation to a UniWed Theory of the Structure of Human
Behavior. Second, revised edition. The Hague and Paris: Mouton.
Prevignano, Carlo L. and di Luzio, Aldo. “A discussion with John J. Gumperz”. This
volume.
Rumsey, Alan. 1990. “Wording, meaning, and linguistic ideology”. American Anthropolo-
gist 92 (2):346–361.
Silverstein, Michael. 1987. “The three faces of ‘function’: Preliminaries to a psychology of
language”. In Social and Functional Approaches to Language and Thought, 17–38.
Orlando and London: Academic Press.
Silverstein, Michael. 1992. “The indeterminacy of contextualization: When is enough
enough?”. In The Contextualization of Language, P. Auer and A. di Luzio (eds.), 55–
76. Amsterdam and Philadelphia: John Benjamins.
Thibault, Paul J. 1986. “Thematic system analysis and the construction of knowledge and
belief in discourse”. In Text, Discourse, and Context: A Social Semiotic Approach.
Victoria University, Toronto: Monographs, Working Papers and Prepublications of
the Toronto Semiotic Circle 3:44–91.
Thibault, Paul J. 1989. “Semantic variation, social heteroglossia, intertextuality: Thematic
and axiological meaning in spoken discourse”. Critical Studies 1 (2):181–209.
Thibault, Paul J. 1990. “Questions of genre and intertextuality in some Australian televi-
sion advertisements”. In The Televised Text, R. Rossini Favretti (ed.), 89–131. Bolo-
gna: Pàtron.
Thibault, Paul J. 1991. Social Semiotics as Praxis. Text, Social Meaning Making and
Nabokov’s ‘Ada’ [Theory and History of Literature Series 74]. Minneapolis and
Oxford: University of Minnesota Press.
Thibault, Paul J. 1994. “Text and/or context?”. In The Semiotic Review of Books (Toronto)
5 (2):10–12.
Thibault, Paul J. 1997. Re-reading Saussure. The Dynamics of Signs in Social Life. London
and New York: Routledge.
Thibault, Paul J. and Van Leeuwen, Theo. 1996. “Grammar, society, and the speech act:
Renewing the connections”. Journal of Pragmatics 25:561–585.
Van Leeuwen, Theo. 1992. “The schoolbook as a multimodal text”. Internationale Schul-
buchforschung 14:35–58.
Wittgenstein, Ludwig. 1969 [1921]. Tractatus Logico-Philosophicus / Logisch-Philoso-
phische Abhandlung. D.F. Pears and B.F. McGuinness (trans.). London: Routledge
and Kegan Paul.
Gumperz and the minims of interaction 63
Chapter 5
On Gumperz and the minims of

interaction1
Carlo L. Prevignano
1. Introduction
The aim of my contribution is to focus on what I call the old problem of

human interaction (“How is human interaction possible?”), and on what
I propose to deWne as the minims of interaction, as distinct from the maxims.
I see Gumperz as a minimist, as compared to Grice, Leech et al., the maximists.
I argue that everyday human solutions to the interaction problem can be
explained by invoking (specifying and analyzing), not only a set of maxims,
but Wrst of all a set of minims, corresponding to minimal tactical requirements
for the application of interaction-types and activity-types. Here I conWne
myself to considering only some of these minims, in particular the ones I
connect to “activity-type cues”. I shall call these activity-type minims. They
work as interaction minims as well. One of them says: “Signal what you are
doing,” or “Signal the activity-type you are performing.” Other activity-type
minims, which function also as interaction minims, say: “Signal the beginning
of your performance according to an activity-type,” “Signal its end,” and
“Signal and conWrm your activity-type involvement.” Minims govern what
Duranti (1985:206V; 1992:48) has called “boundary markers”, that is, con-
ventional cues marking the beginning and closing of an interaction or an
activity-type application, or starting and concluding them tout court.
Protocols for human interaction, as well as for activity-types, include both
minims and maxims, but it is the application of maxims which is constrained
by the application of minims, and not vice versa (minims come Wrst). The
execution of minims by human agents gives cues, or clusters of cues, as
output, which in turn are processed as input by other persons, according to
input minims. In the latter case, minims are used as control and monitoring
goals (see the goal approach in Castelfranchi 1994), or as plan constituents

(see the plan approach in Prevignano 1993). In other words, if the output
minim, “Signal what you are doing”, has, as its corresponding input minim,
something like “Look for what the other is doing,” or “Look for the other’s
activity-type cues,” it is under this input minim (goal or plan) that some cues
as input are tentatively ascribed to an activity-type, thereby triggering the
connected protocol and the included constraints on input maxim application,
as well as activity-speciWc inferences (à la Levinson 1979).
If we agree that activity-type cues “are usually produced and interpreted
without conscious reXection” (to use the words Gumperz has adopted for
“contextualization cues”; see Prevignano and di Luzio, this volume: 14), then
we have to assume that applications also of minims usually occur without
conscious reXection. Usually, but not always. To Grosjean’s remark that “la
non-survenue d’un indice attendu ou sa survenue dans une position non
habituelle vont être interprétées” (2001:147), it is therefore possible to add
that the absence of the cues corresponding to an expected activity type may
indicate an intentional non-application of an output minim or an output-
minim violation (e.g. the intentional violation of the “signal the activity-type
you are performing” minim), and what is interpreted as an intentional output-
minim violation may induce implicatures.
As to minims, here I shall conWne myself to indicating a list of open
questions. Do they correspond to interactional universals in human cultures?
Do the minims of interaction include more than the activity-type minims?
Are there speciWc minims corresponding to cultural and historical interac-
tion-types? What about the relationship between interaction minims and
participation minims?
Regarding human act/action/activity types, while Gumperz has consid-
ered and studied the verbal ones only (what I call the “Gumperz level” in
section 3), I do not exclude, in my present considerations, any possibility,
from entirely and predominantly non-verbal, to predominantly and even
exclusively verbal types (see Table 1). Interestingly, for Gumperz, activities or
activity types now correspond to participants’ or conversationalists’ “cogni-
tive constructs” or “activity constructs” (Prevignano and di Luzio, this vol-
ume: 14), which do not apply to interactions as whole, but to individual
moves, and are characterized as being “subject to constant change in the
course of the exchange” (Prevignano and di Luzio, this volume: 13). If it is
possible to assume that there is usually a constant change in participants’
application of verbal activity types during an exchange (see also Gumperz

1979:274, where everyday conversation is presented as being “marked by
constant transitions from one mode of speaking [or type of verbal activity,
CP] to another; shifts from informal chat to serious discussion, from argu-
ment to humor, or narrative to rapid repartee, etc.”), then activity constructs
seem rather to be considered today by Gumperz as participants’ models for
producing and interpreting moves. I think it is useful to distinguish between
participants’ constructs and analysts’ representational and explanatory con-
structs for those of participants, as well as between act/action/activity tokens
and types. At this point, I propose another distinction between folk or cultural
typologies and tokenologies, on one side, and analysts’ or culturological ones,
including praxeological, action-theoretical and activity-theoretical, of all
kinds and schools, on the other.
A further remark. To be explicit, here I assume not only a 3-, 5-, or 10-level
theory of (joint, we- or collective) activity and interaction, but an n-level one,
and consequently a corresponding n-level analysis.
2. Some preliminaries
After the diVerent interpretations (objectivist, subjectivist, experientialist,

constructivist, dialogic or interactionist, inferential or heuristic, and even
skeptical, see Berge 1994) of communication, communication about com-
munication, communication about and for action and interaction, perhaps it
is now the turn of the analysts — the communication and interaction ana-
lysts — to be analyzed. What type of analyst does Gumperz correspond to?
Pseudo-omniscient? Imperfect? Diabolical? I agree with Levinson (this vol-
ume:38): “cheery optimist”. But what is left over from a cheery optimist’s
analytical game and models of communication and interaction, after the
assault of “radical doubt” (Levinson, this volume:38)? A “postmodernist
conclusion” (again in Levinson’s words), a skeptical idea of both communi-
cation and interaction, of the analytical game itself? I will resist the tempta-
tion of arguing that a skeptical analyst is not self-defeating and what he or
she needs is, precisely, Gumperzian optimism and openness for nourish-
ment. I will take that for granted. But what about a skeptical meta-analyst?
Here I try to play that role.
If communication and interaction analysts, as human believers (for artiWcial

ones see Ballim and Wilks 1991), are today in a postmodern condition — la
condition postmoderne is not yet posthuman! — and their rejection of grands
récits, big theories and macromodels, seems to be a must, their analysis reports
are still récits, i.e. discourses, narrated argumentations. And how can we
reconstruct and assess analysts’ argumentations, their research goals and
procedures? How can we reconstruct their itineraries, starting from their self-
images, pre-theoretical assumptions, questions and doubts, and proceeding to
prediction-making and dis/conWrming, to elicitation and re-elicitation proce-
dures? How can we accomplish all this, if not via their reports? The meta-
analyst has to move between terra incognita and fable convenue (as Koerner
1973 called them, and cf. Prevignano 1979:23–26) in order to reconstruct
analysts’ programs and results.
The Galilean Eppur si muove (“And yet it moves”) here becomes Eppur si
comunica (“And yet we communicate”). However skeptical, a communication
analyst cannot reject this assumption, on pain of self-dismissal. The story is
the same for the meta-analyst. The methodological solipsism of both (together
with Ballim and Wilks’ cognitive solipsism) cannot but be considered meth-
odological, that is, an antidote or a means of defence against reiWcations. And
yet we communicate in communication research as well. Meanwhile, Babel
has turned electronic, and along with the preWx e-, not only do we have e-
communication (as well as e-orality and e-literacy), but also e-miscommuni-
cation and e-misunderstanding. Inevitable misunderstanding (Blum-Kulka
and Weizman 1988)?
Looking for the locus, or the loci, of (pre-electronic) misunderstanding
and misinteraction during intercultural encounters, the Wsh turned ichthyolo-
gist and reached contextualization cues, as well as interactional meanings (i.e.
micro), after he had studied “social meanings” (i.e. macro), and their not
merely “segmental” attachments (see Blom and Gumperz 1972:417–18).
Some of the contextualization cues could now be redeWned as “scripticals” in
relation to activity protocols or scripts. Here I use protocols and scripts
interchangeably. They correspond to both representational and procedural
constructs of conversationalists, who need models of their own activity and
interaction types and tokens, as well as the respective know-how. For their
part, human analysts need sets of assumptions, constraints and procedures as
constructs for representing not only cultural and historical types and tokens
of human activities (from largely unscripted to rigidly scripted ones), but also
the activity models and the know-how which are used by participants them-
selves to categorize and monitor “what’s going on” and to act in life. Following
Gumperz, Svennevig (1999:11–12) prefers to call the latter “activity frames”.
3. The Gumperz level
As an analyst of the “fabric of our social life” (1982a:7), sub specie commu-
nicationis, Gumperz has contributed to the identiWcation of an interaction
principle which I intend to present here in three diVerent speciWcations, each
on a diVerent scale (see level i. to level iii. in Table 1). These range from a
general action-theoretical or praxeological interpretation to a linguistic one.
A caveat. Gumperz has contributed to the individuation of that principle
in the sense that there is a complicity in the crime (and his accomplices
include, at the very least, Bateson, GarWnkel, GoVman, Sacks, Erickson and
Levinson — an invisible college à la Winkin (1981/1995:21), now quite vis-
ible; see Kendon 1990 and Auer 1992:22–23).2 However, it is Gumperz who
has provided its most extensive analyses at the linguistic levels of speciWcation,
in particular studying what he deWnes as “contextualization cues”.
Here is the principle, which I propose to call the “semiotic principle of
interaction”: human agents do not simply act, but are also accustomed to
signaling or cueing the types of act/action/activity they are engaging in; in other
words, they make them interpretable at least via what could be called “Gum-
perzian cues”, rather than (only) with overtly lexicalized or verbalized explana-
tions and rationalizations. Among other reasons, they do so in order to
economically facilitate coordination, to simulate, and to lower or raise the
anxiety level of the other agents present. “Uncertainty reduction” (Giles et al.
1979:372; Berger and Bradac 1982) is accompanied by certainty induction,
and certainty reduction by uncertainty induction. Table 1 presents three
speciWcations of the semiotic principle of interaction.
I cannot resist the temptation of calling the Wrst level of speciWcation of the
principle, the “GarWnkel-and-Sacks level”, the second, the “Levinson level”
(see Levinson 1979), and the third, the “Gumperz level”.3 While presenting
“interpretive sociolinguistics”, Auer and di Luzio have anticipated a deWnition
of what I call the Gumperz level, since they have referred to “the participants’
continuous task and accomplishment to make interpretable and interpret
each other’s linguistic activities” also via cues (1984:viii). A distinct, but
Table 1. Three speciWcations of the semiotic principle of human interaction.

In order to be made intelligible/interpretable/recognizable:
i. all human
ii. entirely and predominantly non- act/action/activity types are
verbal as well as predominantly signaled at least by cues
verbal
iii. even exclusively verbal
parallel, set of tasks (i.e., signaling and identifying or detecting signaled prag-
matic functions of diVerent linguistic activities or activity-types) has been
considered by Stati (1982, 1990) and others from the perspective of dialogue
analysis, and in relation to dialogue constituents other than contextualization
cues. In my view, the Gumperz level corresponds to both cueing itself (as the
process of proposing the use of x as a cue for y as a cued entity, i. e. of a cueing
function and its functors; see Kendon 1996) and cue processing (emission and
reception of both coded and still uncoded or unshared cues), with a distinc-
tion (which has already been used by Richard Petty and John Cacioppo in
their research on persuasion) between central and peripheral processing.
4. A cue/cueing model and analysis
Gumperz has adopted GarWnkel’s program for studying human activities

“through language” (1982a:172), i.e. via verbal signals, and in particular via
what he has termed “cues” (entre autres, in his experiments at Berkeley with
Frederick Erickson et al. on interactional synchrony/asynchrony, see Gum-
perz 1979:278), rather than “markers” (see the markers of Brown and Fraser
1979, as well as of Giles et al. 1979, and the latter’s in nuce “marker [and
marking] theory”). Gumperz begins by working back from participants’ and
other members’ assessments to the surface of the verbal interaction (cf. the
notion of “surface” in Orletti 1994). He has managed to combine a top-down
approach with a bottom-up one, studying both surface constituents and high-
level inferencing, as well as their couplings. In this way, he has provided in
advance a solution to the debate on the predominance/priority of surface
structure vs. deep structure in approaches to human verbal interaction (see
Castelfranchi 1994; Orletti 1994). Yet Gumperz’s surface is populated by more
than just triggers of one-to-one relations (as studied in SchiVrin 1987 or
Bazzanella 1990, 1994; see also Dascalu Jinga and Vanelli 1996), if, impor-
tantly, Gumperzian signals also include clusters of co-occurring cues. As both
output and input, the latter require strategies, not simply processing rules. In
order to capture and represent the many-to-many semiotic circuitry which
Gumperz has characterized as “contextualization” (both in output and in
input), Peircian semiotics is not enough, though this question cannot be
explored here. Gumperzian cues can be coded and uncoded, shared and
unshared, but also plural, as multimodal clusters of co-occurrent triggers.
Further, there is another research problem, which is also pointed out by
Bateson’s metaphor of a shower of signals (1971/1981:125; for Bateson’s cues
and their functions, cf. Rawlins 1987), when these signals are considered as an
output from and an input for human agents. It is the problem of possibility
conditions for their human processing, and in general for the human process-
ing of semiotic multimodality (cf. Thibault, this volume).
If from a “structureless barrage-of-signals view” of human interaction (i.e.
an idea of interaction as an “unstructured exchange” of signals, again in the
words of Brown and Levinson (1979:293)), via a “cross-shower of signals”
view of it (à la Bateson), we come to a structural approach to interaction as a
multichannel event, it is important to avoid studying it only as a turn struc-
ture. An antidote against today’s triumphant reduction of human interaction
to a turn structure and its variants (from a turn-by-turn sequence to a bundle
of n tracks, for n participants, i.e. a multitrack space with tracks being multi-
channel and turns occurring on each of the multichannel tracks; I propose this
variant in Prevignano and Thibault, 2002:166), an antidote, I say, can be an n-
level theory and analysis of interaction, as well as a corresponding n-level
processing model.
In this direction, one of Gumperz’s contributions has been to analyze
occasions of human interaction on a move level, rather than on a turn level,
seeing “actors engaged in strategically formulating and positioning their
moves,” as well as interpreting and reinterpreting other participants’ moves.
Gumperz has distinguished “moves”, “countermoves”, “conWrmations”, and
“disconWrmations”, i.e. move sequences (see Prevignano and di Luzio, this
volume:12). What could be called “move processing” is analyzed by Gumperz
as conversationalists’ execution of “rhetorical strategies” and interpretive pro-
cedures, with positional (etc.) constraints on their applications. Among inter-
actional abilities, it is “rhetorical ability”, as the command or control of a set of
rhetorical strategies being developed in networks of interactions, that
Gumperz (1990b:42) has studied at the move level. A multilevel analysis of a

dissertation defence has been provided by Gumperz (also in collaboration
with his wife, Jenny Cook-Gumperz, and in relation with the so-called Mul-
tiple Analysis Project) in a number of studies (Gumperz 1989b, Cook-Gum-
perz and Gumperz 1994, 1996), where an idea of conversational politics (or, as
it is called, “politics of conversation”) is presented and used to capture a set of
strategies and decisions governing the entire dynamics or history of an inter-
action, and in particular “what can and cannot be said, and […] how and in
what tone information is to be presented” (1996:171).
A further remark. Interestingly, another multilevel approach to talk-in-
interaction is proposed by Catherine Kerbrat-Orecchioni (1997), whose lin-
guistique interactionniste or interactionnelle, starting from the Gumperzian
motto, “speaking is interacting” (Gumperz 1982a:29), reaches entre autres a
sparkling and realistic idea of interaction as “bricolage interactif incessant”
(1998:61), while in interaktionale Linguistik (Selting and Couper-Kuhlen
2001), one of the Gumperzian foci is on conversational or interactional
prosody, e.g. the rhythm and tempo of spoken interaction (Auer, Couper-
Kuhlen and Müller 1999).
Cued entities à la Gumperz — activity-types in this case — require a “large
dose of inferential reconstruction” (Levinson, this volume:37) by both the
participants and the analysts. Moreover, Gumperz-the-analyst’s methodology
is hermeneutical and reconstructive: “The analyst’s interpretation is a recon-
struction of the participants’ interpretation. The assessment of the analyst,
who has to reconstruct, that is, to know and understand, the ethnographic
context [in our case, activity-types, CP] as if he were a participant member, is
a reconstruction of the participants’ assessments, and may possibly be con-
Wrmed by them” (di Luzio, this volume: 6, and cf. Ensink 1987).
In this way, a “singsong rhythm” (Gumperz 1974:11; 1982a:34), that is,
the rhythmic mode of intonation (Gumperz 1995b) of the well-known,
(1) Ahma git me a gig [Gumperz’s rough gloss: I’m going to get
myself some support]
was discovered by Gumperz, working back to the utterance surface from some
participants’ and other members’ assessments, to be a cue accompanying the
utterance which was addressed by a black student to other black students.
Gumperz was able to consult four groups of “judges” (1982a:31–32). Accord-
ing to the members of the last group, it is the rhythm which gives the whole
utterance “a formulaic character” (1982a:34). Elsewhere, Gumperz consid-

ered (1) as an instance of “formulaic phrases identiWable through co-occur-
rent selections of phonological, prosodic, morphological and lexical options”
(1982a:133). In any case, because of its formulaic character, (1) was inter-
preted as corresponding to (what I call) an “activity-type minim”, if it was
glossed as (a) below, to use Gumperz’s own words (cf. Prevignano and di
Luzio, this volume: 20):
(a) “I am doing something that minority people like myself
have to do, get support where I can.”
Other glosses, quoted by Gumperz, were:

(b) “I’m still in control” (Gumperz 1982a:31);
(c) “I am just playing the game as we blacks must do if we are to get along
in a white dominated world” (Gumperz 1982a:31–32).
According to (a), (b), and (c), both the output minim (“Signal via cues what
you are doing”) and the corresponding input minim (“Look for what the other
person is doing, i.e. for his or her activity-type cues”), seem to have been
applied by the student and his addressees, respectively. But can the results of a
reconstructive methodology be considered as a post-factum ‘explanation’ of
cases such as (1)? Postdictions (vs. predictions) presuppose the resolution of
postdictability and postdeterminability problems (Prevignano 1991, 1993).
5. The past and the present
There is an extraordinary game which Gumperz has been playing for more
than forty years: integration. Its risk is eclecticism, and Gumperz knows that
he takes “a somewhat eclectic approach” (1996a:359) to interaction via lan-
guages. From when he was Wfty (in 1972), he has subsequently presented
his research as speech event analysis (Gumperz 1972:17–18; cf. Gumperz
1984:281), conversational analysis of social meaning and sociolinguistic
analysis of conversations (Gumperz and Herasimchuk 1973), sociolinguistics
of interpersonal communication (Gumperz 1974), the interpretive approach
to conversation (Gumperz 1982a:6, 35–7; for a presentation of interpretive
sociolinguistics, as Auer called it, see Auer and di Luzio 1984), and interac-
tional sociolinguistic analysis (Gumperz 1982a:210) or interactional socio-
linguistics (Blom and Gumperz 1972:432; Gumperz 1982a was published as

the Wrst volume of the series Studies in Interactional Sociolinguistics; see also
SchiVrin 1994: ch. 4). Gumperz’s trajectory has cut across the ethnography of
speaking, GarWnkel’s ethnomethodology (Gumperz and Hymes 1972a), Sacks
and SchegloV’s analyses (Gumperz and Hymes 1972b and 1972c, respec-
tively), Grice’s cooperation principle (Gumperz 1990a), GoVman’s interac-
tion order and Levinson’s analysis of activity-types (1979), but not his idea of
interactional intelligence (1995), and may now be seen also as a foundational
contribution to an inferential model of human interaction and to what, with
Grundy (1995:123), we can call an “activity-type theory”, with Thomas
(1995:293), an “activity-type approach”, and with Sarangi (2000:6), an “activ-
ity type analysis” with an “integrated model.” I see the peculiarity of Gum-
perzian socio-cognitivism as consisting in the application of an interpretive
approach, to use Gumperz’s own words, to “the analysis of real-time processes
in face-to-face encounters” (Gumperz 1982a:vii). More than thirty years after
its presentation, this research program continues to be actively pursued and
developed (see also Gumperz 1993, 1995a, 1999, 2001a, 2001b, and Gumperz
and Berenz 1993).
6. The future
If to activity-types there correspond activity scripts, describable as “joint

protocols”, i.e. socio-procedural entities, the already rich toolbox of an ap-
proach to human interaction should include a scriptionary, i.e. a repertoire of
scripts, as both a procedural and representational construct completely diVer-
ent from a script-based lexicon à la Raskin 1981 (Prevignano 1994, 1995). A
scriptionary would be useful for the members of the Turing club, as a proce-
dural way of dealing with cultural sets of human activity-types, but also for
participation analysts (Goodwin and Goodwin 1992; Duranti 1997:280–330),
as a representational way of capturing the “participatory quality” (Duranti
1997:46) of human activities, or what are classiWed as “participatory actions”
(Clark 1999:149). From output and input minims to scriptical cues, from
participants’ strategies to those of analysts, a research agenda for studying
human intelligent interactions and interactional intelligence (see Levinson
1995 and Berger 1997, but also Justine Cassell’s implementations) cannot
forget Gumperz’s inspiring contributions. And here ends my piccolo omaggio
to a grande maestro.
Notes
1. Revised and augmented version of a text previously published as Prevignano 1997. Paul
Thibault’s comments were valuable in the preparation of this version.
2. See also Murray (1994) for a group-sensitive Gumperzologist’s report.
3. See Gumperz 1974:18V, 1977:205V, 1980:119, on a distinction between speech acts and
speech activities; Gumperz 1982b:329V, and Gumperz and Cook-Gumperz 1982:11V, on
a distinction between activities and tasks; see also Gumperz 1989a:78–79, 1990a:439,
1992:43V.
References
Auer, Peter. 1992. “Introduction: John Gumperz’ approach to contextualization”. In The

Contextualization of Language, P. Auer and A. di Luzio (eds.), 1–37. Amsterdam and
Philadelphia: John Benjamins.
Auer, Peter and di Luzio, Aldo. 1984. “Introduction”. In Interpretive Sociolinguistics:
Migrants, Children, Migrant Children, P. Auer and A. di Luzio (eds.), vii-x. Tübingen:
Narr.
Auer, Peter and di Luzio, Aldo (eds.). 1992. The Contextualization of Language. Amster-
dam and Philadelphia: John Benjamins.
Auer, Peter, Couper-Kuhlen, Elizabeth and Müller, Frank. 1999. Language in Time: The
Rhythm and Tempo of Spoken Interaction. Oxford: Oxford University Press.
Ballim, Afzal and Wilks, Yorick. 1991. ArtiWcial Believers: The Ascription of Belief. Hills-
dale, NJ: Lawrence Erlbaum.
Bateson, Gregory. 1971. “Communication”. In The Natural History of an Interview, N.A.
McQuown (ed.). MicroWlm collection of manuscripts on Cultural Anthropology 95
(XV):1–40. Chicago: University of Chicago Library.
Bateson, Gregory. 1981/1995. “Communication”. In La Nouvelle Communication [French
translation of Bateson 1971], Y. Winkin (ed.), 116–144. Paris: Editions du Seuil.
Bazzanella, Carla. 1990. “Phatic connectives as interactional cues in contemporary spoken
Italian”. Journal of Pragmatics 14:629–647.
Bazzanella, Carla. 1994. Le Facce del Parlare: Un Approccio Pragmatico all’Italiano Par-
lato. Firenze: La Nuova Italia.
Berge, Kjell L. 1994. “Communication”. In The Encyclopedia of Language and Linguistics,
R. E. Asher (ed.), 614–620. Oxford: Pergamon Press.
Berger, Charles R. 1997. Planning Strategic Interaction. Mahwah, NJ: Lawrence Erlbaum.
Berger, Charles R. and Bradac, James J. 1982. Language and Social Knowledge: Uncertainty
in Interpersonal Relations. London: Arnold.
Blom, Jan-Petter and Gumperz, John J. 1972. “Social meaning in linguistic structure:
Code-switching in Norway”. In Directions in Sociolinguistics: The Ethnography of
Communication, J.J. Gumperz and D. Hymes (eds.), 407–434. New York: Holt,
Rinehart and Winston.
Blum-Kulka, Shoshana and Weizman, Elda. 1988. “The inevitability of misunderstand-

ings: Discourse ambiguities”. Text 8(3):219–241.
Brown, Penelope and Fraser, Colin. 1979. “Speech as a marker of situation”. In Social
Markers in Speech, K.R. Scherer and H. Giles (eds.), 33–62. Cambridge: Cambridge
University Press.
Brown, Penelope and Levinson, Stephen C. 1979. “Social structure, groups and interac-
tion”. In Social Markers in Speech, K.R. Scherer and H. Giles (eds.), 291–341. Cam-
Castelfranchi, Cristiano. 1994. “Ma non dica idiozie!”: Per un modello delle interazioni
verbali al di là della conversazione”. In Fra Conversazione e Discorso: L’Analisi
dell’Interazione Verbale, F. Orletti (ed.), 143–170. Rome: NIS.
Clark, Herbert H. 1999. “On the origins of conversation”. Verbum XXI (2):147–161.
Dascalu Jinga, Laurentia and Vanelli, Laura. 1996. “Mi raccomando eh!: A pragmatic and
phonetic analysis of the Italian interjection eh”. Lingua e Stile 31 (3):393–431.
di Luzio, Aldo. “Presenting John J. Gumperz”. This volume.
Duranti, Alessandro. 1985. “Sociocultural dimensions of discourse”. In Handbook of
Discourse Analysis 1, T.A. Van Dijk (ed.), 193–230. London: Academic Press.
Duranti, Alessandro. 1992. EtnograWa del Parlare Quotidiano. Rome. La Nuova Italia
ScientiWca.
Duranti, Alessandro. 1997. Linguistic Anthropology. Cambridge: Cambridge University
Press.
Ensink, Titus. 1987. “Interpretative processes in discourse. The approach by John
Gumperz”. Journal of Pragmatics 11:517–531.
Giles, Howard, Scherer, Klaus and Taylor, D.M. 1979. “Speech markers in social interac-
tion”. In Social Markers in Speech, K. Scherer and H. Giles (eds.), 343–381. Cam-
Goodwin, Charles and Goodwin, Marjorie H. 1992. “Context, activity and participation”.
In The Contextualization of Language, P. Auer and A. di Luzio (eds.), 77–99. Amster-
dam and Philadelphia: John Benjamins.
Grosjean, Michèle. 2001. “Verbal et non-verbal dans le langage au travail”. In Langage et
Travail: Communication, Cognition, Action, A. Borzeix and B. Fraenkel (eds.), 143–
166. Paris : CNRS Editions.
Grundy, Peter. 1995. Doing Pragmatics. London: Arnold.
Gumperz, John J. 1972. “Introduction”. In Directions in Sociolinguistics: The Ethnography
of Communication, J.J. Gumperz and D. Hymes (eds.), 1–25. New York: Holt, Rine-
hart and Winston.
Gumperz, John J. 1974. The Sociolinguistics of Interpersonal Communication. Urbino,
Italy: Centro Internazionale di Semiotica e di Linguistica. Working Papers and Pre-
publications 33, serie C.
Gumperz, John J. 1977. “Sociocultural knowledge in conversational inference”. In Lin-
guistics and Anthropology [Georgetown University Round Table on Languages and
Linguistics 1977], M. Saville-Troike (ed.), 191–211. Washington, DC: Georgetown
University Press.
Gumperz, John J. 1979. “The retrieval of sociocultural knowledge in conversation”. Poet-

ics Today 1 (1–2):273–286.
Gumperz, John J. 1980. “The sociolinguistic basis of speech act theory”. Versus 26/
27:101–121.
Gumperz, John J. 1982a. Discourse Strategies. Cambridge: Cambridge University Press.
Gumperz, John J. 1982b. “The linguistic bases of communicative competence”. In Analyz-
ing Discourse: Text and Talk [Georgetown University Round Table on Languages and
Linguistics 1981], D. Tannen (ed.), 323–334. Washington, DC: Georgetown Univer-
sity Press.
Gumperz, John J. (ed.). 1982. Language and Social Identity. Cambridge: Cambridge Uni-
versity Press.
Gumperz, John J. 1984. “Communicative competence revisited”. In Meaning, Form, and
Use in Context: Linguistic Applications, D. SchiVrin (ed.), 278–289. Washington, DC:
Georgetown University Press.
Gumperz, John J. 1989a. “Contextualization cues and metapragmatics: The retrieval of
cultural knowledge”. CLS 25 (2):77–88.
Gumperz, John J. 1989b. “Cadrer et comprendre une politique de la conversation”. In Le
Parler Frais d’Erving GoVman, I. Joseph et al. (eds.), 123–154. Paris: Les Editions de
Minuit.
Gumperz, John J. 1990a. “Conversational cooperation in social perspective”. Parasession
on the Legacy of Grice. Berkeley Linguistic Society 16:429–441.
Gumperz, John J. 1990b. “Theory and method in pluriglossia: The interpretive analysis of
language usage”. In Aspetti Metodologici e Teorici nello Studio del Plurilinguismo nei
Territori dell’Alpe-Adria, L. Spinozzi Monai (ed.), 33–49. Tricesimo: Aviani Editore.
Benjamins.
Gumperz, John J. 1993. “Culture and conversational inference”. In The Role of Theory in
Language Description, W.A. Foley (ed.), 193–214. Berlin and New York: Mouton de
Gruyter.
Gumperz, John J. 1995a. “Mutual inferencing in conversation”. In Mutualities in Dia-
logue, I. Markovà et al. (eds.), 101–123. Cambridge: Cambridge University Press.
Gumperz, John J. 1995b. Personal communication to C.L. Prevignano, Bologna, 31
March.
Gumperz, John J. 1996a. “Introduction to part IV”. In Rethinking Linguistic Relativity, J.J.
Gumperz and S.C. Levinson (eds.), 359–373. Cambridge: Cambridge University
Press.
Gumperz, John J. 1996b. “The linguistic and cultural relativity of conversational infer-
ence”. In Rethinking Linguistic Relativity, J.J. Gumperz and S.C. Levinson (eds.), 374–
406. Cambridge: Cambridge University Press.
Gumperz, John J. 1999. “On interactional sociolinguistic method”. In Talk, Work and
Institutional Order, S. Sarangi and C. Roberts (eds.), 453–471. Berlin and New York:
Mouton de Gruyter.
Gumperz, John J. 2001a. “Contextualization and ideology in intercultural communica-

tion”. In Culture in Communication: Analyses of Intercultural Situations, A. di Luzio,
S. Günthner and F. Orletti (eds.), 35–53. Amsterdam and Philadelphia: John Benja-
mins.
Gumperz, John J. 2001b. “Interactional Sociolinguistics: A personal perspective”. In The
Handbook of Discourse Analysis, D. SchiVrin et al. (eds.), 215–228. Oxford: Basil
Blackwell.
Gumperz, John J. and Berenz, Norine. 1993. “Transcribing conversational exchanges”. In
Talking Data: Transcription and Coding in Discourse Research, J.A. Edwards and
M.D. Lampert (eds.), 91–121. Hillsdale, NJ: Lawrence Erlbaum.
Gumperz, John J. and Cook-Gumperz, Jenny. 1982. “Introduction: Language and the
communication of social identity”. In Language and Social Identity, J.J. Gumperz
(ed.), 1–21. Cambridge: Cambridge University Press.
Gumperz, John J. and Cook-Gumperz, Jenny. 1994. “The politics of a conversation:
Conversational inference in discussion”. In What’s Going On Here? Complementary
Studies of Professional Talk, A. Grimshaw et al., 373–394. Norwood, NJ: Ablex.
Gumperz John J. and Cook-Gumperz, Jenny. 1996. “Treacherous words: Gender and
power in academic assessment”. Folia Linguistica XXX (3–4):167–188.
Gumperz, John J. and Herasimchuk, Eleanor. 1973. “The conversational analysis of social
meaning: A study of classroom interaction”. In Sociolinguistics: Current Trends and
Prospects [Georgetown University Round Table on Languages and Linguistics 1972],
R.W. Shuy (ed.), 99–134. Washington, DC: Georgetown University Press.
Gumperz, John J. and Hymes, Dell. 1972a. [“Note to H. GarWnkel’s ‘Remarks on ethno-
methodology’”]. In Directions in Sociolinguistics: The Ethnography of Communica-
tion, J.J. Gumperz and D. Hymes (eds.), 301–309. New York: Holt, Rinehart and
Winston.
Gumperz, John J. and Hymes, Dell. 1972b. [“Note to H. Sacks’ ‘On the analyzability of
stories by children’”]. In Directions in Sociolinguistics: The Ethnography of Communi-
cation, J.J. Gumperz and D. Hymes (eds.), 325–329. New York: Holt, Rinehart and
Winston.
Gumperz, John J. and Hymes, Dell. 1972c. [“Note to E.A. SchegloV’s ‘Sequencing in
conversational openings’”]. In Directions in Sociolinguistics: The Ethnography of
Communication, J.J. Gumperz and D. Hymes (eds.), 346–348. New York: Holt,
Rinehart and Winston.
Gumperz, John J. and Hymes, Dell (eds.). 1972. Directions in Sociolinguistics: The Ethnog-
raphy of Communication. New York: Holt, Rinehart and Winston.
Gumperz, John J. and Levinson, Stephen C. 1996. “Introduction: Linguistic relativity re-
examined”. In Rethinking Linguistic Relativity, J.J. Gumperz and S.C. Levinson (eds.),
1–8. Cambridge: Cambridge University Press.
Gumperz, John J. and Levinson, Stephen C. (eds.). 1996. Rethinking Linguistic Relativity.
Kendon, Adam. 1990. “Some context for Context Analysis: A view of the origins of
structural studies of face-to-face interaction.” In Conducting Interaction: Patterns of
Behavior in Focused Encounters, 15–49. Cambridge: Cambridge University Press.
Kendon, Adam. 1996. “Cues of context” [Review article of The Contextualization of

Language, P. Auer and A. di Luzio (eds.). Amsterdam and Philadelphia: John Ben-
jamins, 1992]. Semiotica 109 (3/4):349–356.
Kerbrat-Orecchioni, Catherine. 1997. “A multilevel approach in the study of talk-in-
interaction”. Pragmatics 7 (1):1–20.
Kerbrat-Orecchioni, Catherine. 1998. “La notion d’interaction en linguistique: Origines,
apports, bilan”. Langue Française 117:51–67.
Koerner, E.K.F. 1973. Ferdinand de Saussure: Origin and Development of his Linguistic
Thought in Western Studies of Language. A Contribution to the History and Theory of
Linguistics. Braunschweig: Vieweg.
Levinson, Stephen C. 1979. “Activity types and language”. Linguistics 17:365–399.
Levinson, Stephen C. 1995. “Interactional biases in human thinking”. In Social Intelligence
and Interaction, E.N. Goody (ed.), 221–260. Cambridge: Cambridge University Press.
Levinson, Stephen C. “Contextualizing ‘contextualization cues’”. This volume.
Murray, Stephen O. 1994. Theory Groups and the Study of Language in North America:
A Social History. Amsterdam and Philadelphia: John Benjamins.
Orletti, Franca. 1994. “Sulla superWcie del conXitto”. In Fra Conversazione e Discorso:
L’Analisi dell’Interazione Verbale, F. Orletti (ed.), 171–184. Rome: NIS.
Orletti, Franca (ed.). 1994. Fra Conversazione e Discorso: L’Analisi dell’Interazione Ver-
bale. Rome: NIS.
Prevignano, Carlo. 1979. “Una tradizione scientiWca slava tra linguistica e culturologia”.
In La Semiotica nei Paesi Slavi: Programmi, Problemi, Analisi, C. Prevignano (ed.),
23–99. Milan: Feltrinelli.
Prevignano, Carlo. 1991. “Dopo la ‘conclusione deprimente’ di Brown e Yule, come
studiare ponti pragmatici in dialoghi?”. In Dialoganalyse III, S. Stati et al. (eds.), 211–
225. Tübingen: Niemeyer.
Prevignano, Carlo. 1993. “On reconstructions of dialogues as interplanning cases”. In
Dialoganalyse IV, H. LöZer (ed.), 183–188. Tübingen: Niemeyer.
Prevignano, Carlo. 1994. “From scripts to interscripting in dialogue analysis”. Paper
presented at the 5th IADA Congress on Nouvelles Perspectives dans l’Analyse de
l’Interaction Verbale. Paris, Université de la Sorbonne Nouvelle, 17–19.3.1994.
Prevignano, Carlo. 1995. “Cross-framing in dialogue”. Paper presented at the 1995 IADA
Round Table on Dialogue Analysis: Units, Relations and Strategies Beyond the Sen-
tence. Bologna, Department of Linguistic and Oriental Studies, University of Bolog-
na, 31.3–1.4.1995.
Prevignano, Carlo L. 1997. “Gumperz and the minims of interaction”. In Discussing
Communication Analysis 1: John J. Gumperz, S. Eerdmans, C. Prevignano, and P.J.
Thibault (eds.), 48–56. Lausanne: Beta Press.
volume.
Prevignano, Carlo L. and Thibault, Paul J. 2003 “Continuing the interview with Emanuel
A. SchegloV”. In Discussing Conversation Analysis: The Work of Emanuel A. SchegloV.
C.L. Prevignano and P.J. Thibault (eds.), 165–171 Amsterdam and Philadelphia: John
Benjamins.
Raskin, Victor. 1981. “Script-based lexicon”. Quaderni di Semantica II (1):25–34.

Rawlins, William K. 1987. “Gregory Bateson and the composition of human communica-
tion”. Research on Language and Social Interaction 20:53–77.
Sarangi, Srikant. 2000. “Activity types, discourse types and interactional hybridity: The
case of genetic counselling”. In Discourse and Social Life, S. Sarangi and M. Coulthard
(eds.), 1–27. Harlow: Pearson Education Limited.
Scherer, Klaus and Giles, Howard. (eds.). 1979. Social Markers in Speech. Cambridge:
Cambridge University Press.
SchiVrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press.
SchiVrin, Deborah. 1994. Approaches to Discourse. Oxford: Blackwell.
Selting, Margret and Couper-Kuhlen, Elizabeth. 2001. “Forschungsprogramm ‘Interak-
tionale Linguistik’”. Linguistische Berichte 187:257–287.
Stati, Sorin. 1982. Il Dialogo: Considerazioni di Linguistica Pragmatica. Naples: Liguori.
Stati, Sorin. 1990. Le Transphrastique. Paris: PUF.
Svennevig, Jan. 1999. Getting Acquainted in Conversation: A Study of Initial Interactions.
Thibault, Paul J. “Contextualization and social meaning-making practices”. This volume.
Thomas, Jenny. 1995. Meaning in Interaction: An Introduction to Pragmatics. London:
Longman.
Winkin, Yves. 1981/1995. “Présentation générale”. In La Nouvelle Communication, Y.
Winkin (ed.), 11–109. Paris: Editions du Seuil.
Winkin, Yves (ed.). 1981/1995. La Nouvelle Communication. Paris: Editions du Seuil.
Commentary on a discussion with John J. Gumperz 79
Chapter 6
A commentary on a discussion
with John J. Gumperz1
Afzal Ballim
My contribution to this volume is that of a non-linguist. At the same time, it is

that of a non-linguist who has worked for many years with computational
linguists, and who works in the Weld of Natural Language Processing (NLP) —
the treatment by computer of natural language. As such, I hope to be able to
comment on the discussion with Gumperz from two perspectives:
1. In what way may Gumperz’s work in sociolinguistics and the ethnography
of communication be of beneWt to those of us working in the computer
treatment of language?
2. How may our work aid, or eVect, that of Gumperz?
I have worked in the Weld broadly known as Natural Language Processing for
over a decade, and during that time I have worked with both computational
linguists and people who, like myself, have an interest in linguistics and
linguistic theories, yet whose concern with computers using language has led
them down diVerent paths from those explored by linguists. This has
prompted me to ask myself why the work of linguists does not have a greater
eVect on work within NLP. After all, we are both concerned with the same
fundamental thing: language. Surely, with this common point of interest,
there should be more interaction, in particular between NLP researchers and
computational linguists who have computational models as an extra point of
commonality?
I believe the reason for the minimal interaction between the two is due to
the fact that:
a. NLP researchers believe that the work of computational linguists, rooted as

it is in providing theoretical models of language, tends in many instances to be
too far from the practical problems of using language for it to have much
bearing on what they do;
b. Computational linguists believe that the work of NLP researchers often

lacks the rigorous theoretical models that they feel are necessary to explicate
language.
Of course, interaction does take place, and the work in computational linguis-
tics on phonology, morphology, part-of-speech tagging, and syntactical anal-
ysis plays an important role in modern NLP. It is when one goes beyond these
realms to semantics and pragmatics that the interaction breaks down. In part,
what I wish to say in this commentary is that the work of Gumperz is an
example of why the interaction should not fail.
In order to situate my detailed comments on the discussion, it is necessary
to give a brief description of my own work in NLP. I am concerned with all
forms of computer-mediated communication; in particular, with human-
machine interaction (and in this I include language-based human-machine
communication), as well as with human-human communication assisted by
computer (such as in Computer-Supported Collaborative Work [CSCW]). The
ultimate goal in this work is to create systems that communicate.
As such, I am less encumbered by having to provide neat theoretical
accounts of linguistic phenomena than are my colleagues in computational
linguistics (although such neat theories are still desirable). Instead, I can take
shortcuts around such philosophical problems as “symbol grounding” or even
“what is semantics?”. Indeed, in recent years my work has concentrated on
“robust interaction”, in which we admit that the computer interpretation and
model of the interaction is incomplete, but we try to work with what we know
the computer can do. This tendency towards robustness and practicality has
been notable in work on human-computer interaction since the mid-90s.
If one can describe the desired functionality of a system, and the system is
able to communicate with a human to achieve that functionality, then the goal
has been reached and it is not necessary to ask if the computer system has
understood. That said, the sheer complexity of the problem of achieving
communication between a computer system and a human is such that one
cannot help but formulate theories that may well be applicable to human-
human communication as well. For the same reasons, theories of human-
human communication must be considered, as they are still important.
The crux of my work (see Ballim and Wilks 1991; Ballim and Pallotta,
forthcoming) is in interpreting utterances in the light of their context as
perceived by the speakers and hearers involved. That is, the system formulates
a model of the knowledge, beliefs, desires, intentions, and other attitudes of
the interlocutors that are relevant in the context of the dialogue, and further-
more it formulates models of how the interlocutors model each other at that
point. These models are then used as the basis for interpretation of the dis-
course. They are built using past and current experience with the individual
interlocutors, as well as through generalizations about the background of
various participants.
In general terms, I feel that this approach is very compatible with Gum-
perz’s ethnographically-based sociolinguistic analysis of dialogue, since no-
tions such as “networks of relationships”, “situated inference”, and “speech
events” are all fundamental to the processing involved. In turn, the eVec-
tiveness of this NLP method can be seen as providing indirect evidence of the
validity of Gumperz’s approach. This can be seen more clearly by considering
in more detail speciWc parts of the discussion.
Indirect reasoning processes
This notion is fundamental to my approach to discourse, by which discourse

must be interpreted with respect to the underlying reasoning processes that
the participants in a dialogue mutually believe each other to use. Peers would
normally believe each other to use similar reasoning, hence their communica-
tion would tend not to explicitly emphasize those processes. In this way,
communication between children would be explained by reference to the
types of reasoning that children would be expected to use. This, in turn, would
allow for hypothesis formation about what aspects we would expect to see
explicitly lexicalized, and what aspects we would expect to be implicit. Given
hypotheses about the reasoning processes, my system could be used as a
predictive model for peer communication, which could be used as veriWcation
of these hypotheses.
Speech events
Gumperz insists that ethnographically-based sociolinguistic analysis must

focus on speciWc speech events. In other words, it is not suYcient to consider
discourse in terms of a uniform, disembodied community. There is no deny-
ing, however, that this notion of a community has got a role to play, but that
role must be mediated by the personal experiences of the individuals involved.
Thus an individual may display a discourse behavior consistent with a par-
ticular ethnic make-up when conversing with peers, yet display a diVerent
behavior when involved in discourse with people from other backgrounds.
This notion of interactively deWned behavior is also a key concept in my own
approach to discourse, in which dialogue is conditioned by interlocutors’
perceptions of each other and of the context.
Context and communication
The notion of the context of a communication has long been central in NLP.
My own work with Yorick Wilks, as well as that of people such as Barbara
Grosz, and Barwise and Perry’s (1983) “situation semantics”, are all examples
of taking context as being of primary importance in discourse understanding.
However, Gumperz’s notion of “contextualization cues” is not something
accounted for or used by any of these theories. His aim to show the interaction
of indexical signs with symbolic signs may, I believe, make a great contribu-
tion to NLP, where the problems of deWning the structure of discourse and
demarcating shifts in context are far from resolved.
Gumperz presents a schema for overcoming some of the ambiguity inher-
ent in the inferential process associated with discourse. I believe that combin-
ing this schema with the methods commonly used in NLP, and those adopted
in my own work (of Wnding hypotheses that are maximally consistent with the
context and interlocutors’ background) may yield an interesting and novel
approach to problems of ambiguity.
The method that I mention in my work of Wnding maximally consistent
hypotheses that explain discourse events has undercurrents of Gumperz’s
“activity types”, in particular because they may be based on predictive reason-
ing on where the conversation is leading the participants. Such predictive
structures are of course a notable aspect of much NLP work (in particular,
Schank’s work on scripts and the derivatives from that work can be seen as
aiming to embody form-context relationships).
Corpus pragmatics
Gumperz is questioned about the use of corpora. His problem with such
corpora is that these transcriptions lack many of the contextual cues that he
believes to be fundamental to understanding conversation. A large part of the
problem is the lack of standards for recording such cues. Given the growing
and already widespread use of corpora in computational linguistics and NLP,
a contribution in this domain which establishes standards for those cues
would be of great beneWt to many people.
Conclusion
In these few paragraphs, I have striven to give an appreciation of Professor

Gumperz’s work from a diVerent perspective, that of someone working on
methods to enable the use of language as a communicative medium between
humans and computers. I hope that the comments have shown that, for myself
at least, there is a lot to be learned from Gumperz’s research by people working
in NLP, and that, reciprocally, work in NLP can be of beneWt to those working
in areas such as sociolinguistics.2
Work on applied computational linguistics (sic) in recent years has shown
that people are taking greater interest in using linguistic theories in the very
practical process of making a computer interact intelligently with humans. It
remains to be seen, however, whether we will be able to move away from using
small parts of linguistic theories taken piece-meal, towards a greater linguistic
foundation for our interactive systems of the future.
Notes
1. This paper was revised in February 2002.

2. See Coakes et al. 2000, Goddard 1998, Greif 1998, Horton and Keysar 1996, Lyons 1977,
and Sproull & Kiesler 1988 for further relevant work in the areas of semantics, pragmatics,
Computer-Supported Collaborative Work (CSCW), and NLP.
References
Ballim, Afzal and Pallotta, Vincenzo. Forthcoming. Dialogue through the ViewFinder.
Ballim, Afzal and Wilks, Yorick. 1991. ArtiWcial Believers: The Ascription of Belief. Hills-
dale, NJ: Lawrence Erlbaum.
Barwise, Jon and Perry, John. 1983. Situations and Attitudes. Cambridge, Mass.: The MIT
Press.
Coakes, Elayne, Willis, Dianne and Lloyd-Jones, Raymond (eds.). 2000. The New Socio-
tech: GraYti on the Long Wall. London: Springer.
Goddard, CliV. 1998. Semantic Analysis: A Practical Introduction. Oxford: Oxford Uni-
versity Press.
Greif, Irene. 1998. Computer-Supported Cooperative Work: A Book of Readings. San
Mateo, CA: Morgan Kaufmann.
Horton, William S. and Keysar, Boaz. 1996. “When do speakers take into account com-
mon ground?” Cognition 59:91–117.
Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press.
Sproull, Lee and Kiesler, Sara. 1988. “Reducing social context cues: Electronic mail in
organizational communication”. In Computer-Supported Cooperative Work: A Book
of Readings, I. Greif (ed.), 683–712. San Mateo, CA: Morgan Kaufmann.
John J. Gumperz: A review 85
Chapter 7
A review of John J. Gumperz’s

current contributions to
Interactional Sociolinguistics1
Susan L. Eerdmans
1. Introduction
In this chapter, I would like to outline some of John Gumperz’s most impor-
tant contributions to Interactional Sociolinguistics in the last few years. An
attempt will be made to deWne in a nutshell Gumperz’s particular analytic
approach, and to demonstrate, with examples taken from two of his case
studies, how this is applied to real data.
‘Interactional Sociolinguistics’ is a general term used to describe a number
of qualitative approaches to the analysis of interactive and dialogic processes
in interpersonal communication. Its roots lie in the Welds of sociology, anthro-
pology and linguistics, as reXected by interactional sociolinguists’ interest in
the closely-knit interplay existing between language, society and culture
(SchiVrin 1994, 1996), and in the “communicative import of diversity”
(Gumperz 1999b:453). The focus of their attention is on naturally-occurring
utterances in social context, the aim of their analysis being to discover and
understand “how interpretation and interaction are based upon the interrela-
tionship of social and linguistic meanings” (SchiVrin 1994:8).
2. Background
As a linguistic anthropologist, John Gumperz has contributed to interactional

approaches to discourse in a signiWcant way, “providing both the empirical
evidence and the analytical framework for investigating the varied but system-
atic ways in which language shift both reXects and deWnes social and cultural
boundaries” (Gumperz 1992a:229). By focussing his sociolinguistic analysis
on the qualitative aspects of verbal and non-verbal processes, together with
those of shared cognition and understanding, Gumperz has demonstrated his
commitment to a general theory. This is not only a theory of language, but of
verbal communication, of communicative action, “which integrates what we
know about grammar, culture and intercultural conventions into a single,
overall framework of concepts and analytical procedures” (Gumperz 1982:4).
The analytical methodology adopted by Gumperz for his in-depth investiga-
tion of discursive processes in what GarWnkel calls “naturally organized activi-
ties” is based principally on the inductive procedure of analysis adhered to by
conversation analysts. 2 The aim of the conversation analytic, turn-by-turn
study of sequential processes is to discover the naturally-occurring and recur-
ring mechanisms and patterns of conversational management, seeking “to
describe the underlying social organization[...]through which orderly and
intelligible social interaction is possible” (Goodwin and Heritage 1990:283).3
In keeping with conversation analysts’ ethnomethodological background,
which leads them to avoid any premature generalization or idealization, this
research is carried out assuming very little about participants’ social and
cultural background a priori; aspects of the context such as the social identity
of participants and the setting are considered to be “a category of social life
and conduct that is subject to locally situated interpretive activity” (SchiVrin
1994:235).
Alongside elements of the conversation analytic method of investigating
sequential processes, Gumperz’s interactional sociolinguistics also employs
features of more general approaches to discourse analysis, including, for
example, Gricean analysis of inferential processes and implicature. One of the
key notions of Gumperz’s perspective on verbal communication is communi-
cative intent, the interpretation made by the listener(s) of what the speaker is
trying to communicate, roughly correspondent to the illocutionary force of
what is conveyed (Gumperz 1992a:230).4 The assessment of participants’
communicative intentions is an essential part of conversational inference,
another fundamental concept that Gumperz is concerned with, deWned as
“the situated and presupposition-bound interpretive processes by which in-
terlocutors assess what they perceive at any one point in a verbal encounter
and on which they base their responses” (Gumperz 1996c:375). In the context
of a “communicative ecology” (Gumperz 1990), conversational inference thus
corresponds to the interactionally-based process by which speakers and lis-

teners cooperate to produce and evaluate speech.
According to Gumperz, this process of conversational inferencing im-
plies a three-part analysis of meaning and interpretation, in that at least
three speaking turns are involved: move, counter-move and conWrmation or
disconWrmation.5 One of the constituent moves of such a three-part ex-
change may also be non-linguistic, as illustrated by GoVman in the follow-
ing example:
A: [Enters wearing a new hat]

B: No, I don’t like it.
A: Now I know it’s right.
(GoVman 1976:290, cited in Tsui 1989:547)
In his current research, Gumperz’s basic assumption about language owes a

great deal to work in metapragmatics carried out mainly in the United States
and particularly by Michael Silverstein and his students.6 This approach to the
analysis of verbal communication is of a semiotic functional nature, its key
concept being the Peircian notion of indexicality and the distinction that
Peirce makes between symbolic and indexical signs. By symbolic signs, Peirce
basically refers to arbitrary and conventional signs which denote their object
on the basis of a general law or convention. For Gumperz, symbolic signs,
which he describes as “denotational” (Gumperz and Toribio 1999), include
the information that is conventionally conveyed via lexical meaning and
grammatical processes. With indexical signs, on the other hand, the signaling
value rests on the conventional associations between particular kinds of in-
dexes or signs and context, this being a direct linkage, in no way mediated by
any symbolic process. Symbolic signs therefore communicate by means
of grammatical and lexical rules, whereas indexical signs communicate “by
virtue of direct conventionalized associations between signs and context es-
tablished or transmitted through previous communicative experience” (Pre-
vignano and di Luzio, this volume: 10).
We can further distinguish between two main types of indexical signs: (1)
well-known, referential indexes (in Jakobson’s terms, “shifters”), such as
deixis, anaphora and other discourse markers, which have at least some
propositional content, some kind of semantic component. These indexical
signs refer directly to context and can therefore be explained only in terms of
how they are used in context, or, as Gumperz argues, in terms of how they
enter into a particular interactive or dialogic process; (2) what Silverstein calls
“pure indexes” whose signaling value is basically relational, indexical, in that
they possess no propositional content. These indexical signs refer to, or rather,
index a context; this signaling process does not depend on any kind of logical
entailment, but on “conventionalised associations between co-occurring clus-
ters of signs and contextual frames” (Gumperz 1995:118).
One set of pure indexical signs Gumperz is particularly concerned with is
that of contextualization cues, “verbal and non-verbal metalinguistic signs
that serve to retrieve the context-bound presuppositions in terms of which
component messages are interpreted” (Gumperz 1996c:379). These cues op-
erate principally at the level of prosody, paralinguistic signs, code choice and
choice of lexical expressions or formulaic sequences (Gumperz 1993), their
meaning being established through interaction and not by the assignment of
context-free semantic content. In this way, they convey limitations on inter-
pretation, for example by cueing “the implicatures or lines of practical reason-
ing that enter into the interpretation of a particular word or expression”
(Gumperz 1992b:50).
The background knowledge necessary for conversational inference is
made available by means of such contextualization cues, which “when pro-
cessed in co-occurrence with symbolic grammatical and lexical signs [serve]
to construct the contextual ground for situated interpretation, and thereby
[aVect] how constituent messages are understood” (Gumperz 1999b:461).
Gumperz’s approach to the analysis of discursive practices is very much
focussed on how this interaction between indexically-linked signals, on the
one hand, and symbolic signs, together with the sequential organization
of speech exchanges, cultural and other background information, on
the other, forms everyday social action, that is to say, goal-oriented commu-
nicative action.
3. Research method
To illustrate the methodological procedures adopted by Gumperz when car-

rying out his research, examples will be taken from two particular case studies,
one concerning a job center interview and the other a rape trial cross-exami-
nation.7 The advantage of using naturally-organized situations of this kind is
that they are in a sense socioculturally-deWned, in that we have a relatively

clear idea of what cultural presuppositions they involve. In other words, we
possess typiWed knowledge about such events, these typiWcations being cultur-
ally based: that is, they are built up and acquired over time through “histori-
cally speciWc interactive experience” (Gumperz 1992b:45). 8 By ‘culture’ here,
Gumperz refers to cultural knowledge in the linguistic sense of the term, that
is, the knowledge that we need, that we rely on, to communicate.9
The starting point of Gumperz’s interactional investigation is ethno-
graphic research.10 Whereas the second case study discussed here, the rape
trial cross-examination, is a re-analysis of a study carried out by Paul Drew
(Drew 1992), in which Gumperz’s main aim is to demonstrate the relative
inadequacy of sequential analysis in accounting for situated interpretation,
the Wrst example, the job center interview, is a clear exempliWcation of Gum-
perz’s ethnographic approach.
The data for the latter case study were collected during a period of about
three years’ research on intercultural communication, mainly carried out in
Great Britain. In this research study, Gumperz began his work as ethnogra-
pher with a period of interviewing in order to discover people’s opinions
about their communication with people from diVerent ethnic backgrounds, in
this case, Asians and Cypriots in London and the Midlands. Like most ethnog-
raphers, Gumperz started his investigation by asking a lot of seemingly ran-
dom questions, searching for a hypothesis that would lead to a particular kind
of query: “What is it about intercultural communication that people see as
signiWcant?”. He was thus able to start to get a feel of the situation and to
obtain a great deal of ethnographic information on “the history, social charac-
teristics, everyday practices, and goals” of the community concerned (Gum-
perz and Berenz 1993:92).
One of the most important elements of ethnographic Weldwork is partici-
pant observation in the community so that the researcher can begin to absorb
the atmosphere and familiarize him/herself with the communicative ecology
of the setting. Through contacts with the National Centre for Industrial Lan-
guage Training, which uses ethnographic approaches to intercultural commu-
nication in order to diagnose, for example, problems on the shop Xoor,
Gumperz was in fact able to participate as observer in various work situations,
including a cafeteria at Heathrow Airport.11 In this way, he could observe for
himself the occurrences of conversational misunderstanding and breakdown
and obtain Wrst-hand information on the dynamics of the intercultural verbal
and non-verbal interaction taking place.

In interactional studies of this kind, audio and/or video recordings of
varying lengths are made of the encounters most likely to prove useful for
testing the ethnographer’s hypotheses, and this recorded material is then
analyzed at the level of content and rhythmic organization in order to identify
thematically coherent and sequentially-bounded units, that is, speech events.
These can be deWned as “temporally organized sequences of verbal acts, sepa-
rable from the surrounding talk by empirically identiWable beginnings and
ends or outcomes, which provide evidence with which to validate the analysis
of constituent interpretive processes” (Gumperz 1996c:380).
The next step in the analytic procedure is to make detailed transcriptions
of the isolated event sequences in order to prepare interactional texts and
analyze them in greater depth. 12 The aim of this analysis is to investigate the
interactive features of conversational management and to discover “the per-
ceptual cues that speakers and audiences rely on in signaling what they intend
to convey and in assessing what they hear” as part of conversational inference
(Gumperz and Berenz 1993:119). These perceptual cues can be verbal or non-
verbal and include paralinguistic, prosodic, formulaic and other contextua-
lization cues such as style-shifting and code-switching.
3.1 Job center interview
The Wrst case study to be examined here, the job center interview, reXects very
clearly Gumperz’s ever-present interest in intercultural communication, and
demonstrates the extent to which interactional sociolinguistic analysis can
reveal the stated and unstated presuppositions that enter into interactions of
this kind.
As mentioned previously, the interview, aimed at obtaining information
about the candidate’s past work experience, was recorded in the Midlands as
part of a three-year project on intercultural communication. The interviewer
was a native speaker of the English spoken in that area, and the interviewee a
South Asian living locally.13 As one of a number of such situations, “compared
systematically with a range of other encounters both within and outside the
particular institutional context” (Gumperz and Berenz 1993:93), the analysis
of this interview reveals the interviewer’s preoccupation with Wnding out
exactly what the candidate does, the actual work he performs. With this
knowledge, she is able to judge the kind of technical skills involved and
consequently obtain detailed information about the person as an individual.

The South Asian candidate, however, consistently disappoints the inter-
viewer by giving only general descriptions of his work program. Although he
correctly interprets her responses (“yeah”, “hmm” etc.) as expressions of
dissatisfaction and therefore as prompts eliciting more details, he does not
have a clear idea of what she is looking for. Even when, more and more
impatient, she uses the prosodic strategy of accenting to focus her questions
on him as an individual, “what do you actually *do?”, “how involved are
*you?”, he still gives a general answer, “he shows us what to do.”14 An obvious
eVect of a sequence of nonspeciWc responses of this kind is that it gives the
impression of noncooperation, and can therefore lead to negative attitudinal
judgements by the person conducting the interview.
Gumperz’s argument is that some of the responses the interviewer re-
ceives from the candidate are actually generated by the interaction itself. In
fact, one of the most striking features of the dialogue is that at no point does
the interviewer really make clear exactly what she is interested in. She is never
very speciWc lexically and only indicates the extent of her dissatisfaction
indirectly by modulating the prosody of her single word responses or by
accenting key words. Not surprisingly, these indirect strategies used by the
native speaker to elicit more detailed information are not transparent enough
for the candidate; even though he realizes there is some problem in the
communication, he seems “unable to process the interviewer’s use of stress as
a guide to indicate what type of an answer is wanted” (Gumperz 1992c:323).15
The communicative diYculty demonstrated by this case does not seem to
be due to lack of formal linguistic knowledge of English on the part of the
interviewee, whose grammatical competence is fairly good. The problem in-
stead appears to arise from the fact that his interactive experience of the
language has been primarily with other immigrants from South Asia living in
his neighborhood. These people have no problems at all when using English
interactively amongst themselves, but they sometimes run into trouble when
communicating with native speakers, due to their tendency to map the pro-
sodic and rhetorical strategies of their own languages onto English.16 One of
the characteristics of this mapping process is that South Asian immigrants like
the interviewee do not use stress placement pragmatically when adopting
English in dialogic interaction, relying on “phonological means for marking
prosodic distinctions and relations” which diVer from those of native speakers
of English (Gumperz 1982:122).17 As a consequence, the participants of inter-
cultural encounters of this kind frequently experience serious interactional

problems, which nearly always imply reciprocal misunderstandings and, in
some cases, even breakdowns in communication.
Together with these pragmatic causes for native-nonnative misunder-
standings, also higher level causes exist, functioning at the level of interactive
norms.18 By re-examining transcripts of interviews similar to the one de-
scribed above, this time focussing his attention on the content of the encoun-
ter, Gumperz is in fact able to give a further explanation of what is happening
in the interaction and to start to understand what type of “South Asian activity
level presuppositions” are involved (Gumperz 1996c:400). One of his Wrst
Wndings is that there is a systematic avoidance on the part of the immigrant
interviewees to talk about themselves; this is clearly theme-speciWc, in that
they tend to avoid revealing information which could be considered an ex-
pression of what they feel or believe about certain matters. Another important
pattern to emerge from Gumperz’s analysis of topic is that the South Asian
speakers of English participating in these interviews again systematically
avoid presenting information that they assume their interviewers already
know. For example, at the beginning of a face-to-face interview, when asked
where they live (the interviewers simply checking the information they have
on the form in front of them), the interviewees nearly always give very vague
responses, demonstrating their unease by mumbling or stuttering.
There thus seems to be a systematic diVerence in the ideology of intercul-
tural relations that governs the conduct in such situations. All over South
Asia, social encounters such as these job interviews are seen as “relatively
formal hierarchical situations of pleading” (Gumperz 1996c:399, my empha-
sis), in which, for the context of that particular situation, the interviewee
considers himself as hierarchically inferior and the interviewer is regarded as
the expert. Volunteering personal information, especially information the
interviewer is believed to already possess, is thought to be presumptuous; this
explains the systematic stammering and general hesitation characterizing the
responses of South Asian interviewees when forced to talk about themselves.
Connected to the latter sociolinguistic behavior is the tendency of these immi-
grants to de-emphasize their qualiWcations, to “background their own accom-
plishments” (ibid.).19 By letting the interviewer discover this information
from the records, rather than provide it themselves and risk appearing boast-
ful, they show that what they are actually doing is pleading with their ‘supe-
rior’ to represent their case as an individual.
3.1.1 Concluding remarks

In conclusion, it appears to be the case that in intercultural encounters of this
kind, serious communicative problems can arise due to ideological diVer-
ences, to diVerent ways of viewing or framing such situations on the part of
South Asian speakers of English, the term frame being used by Gumperz to
indicate “a class of related presuppositions that guide conduct in certain
situations and that are directly related to choice of topic and verbal strategies”
(Prevignano and di Luzio, this volume: 27). More speciWcally, it is the interac-
tion taking place between inferences both at an activity and a sequential level,
together with “the workings of contextualization conventions” and the gen-
eral tendency of people to base interpretation on their own culture-bound
presuppositions, that may lead to diYculties in communication and possible
breakdowns in conversational cooperation (Gumperz 1996c:400).
In situations of interethnic miscommunication, the interplay between the
various potentially conXicting aspects indicated above tends to result in what
Gumperz calls a “compounding” of interactional diYculties: misunderstand-
ings occurring at one level may get repaired, but very often this act of repair is
only partially successful or fails completely, thus compounding the communi-
cative problems by adding yet another element of diYculty to an already
complicated situation. Such a build-up in communicative tension, not only
from the case at hand, but also from cumulative experience of similar situa-
tions in the past which may have favored the development of stereotypes,
inXuences the interviewer’s “pejorative evaluation” of the candidate’s perfor-
mance (Gumperz 1996c:400). Problems which in other circumstances would
probably be interpreted as examples of lack of linguistic knowledge, in the
hierarchical context of the job interview are considered to reXect “the
speaker’s ability, truthfulness or trustworthiness” (Gumperz 1992c:326–327).
On the other hand, also the interviewee is bound to feel dissatisWed with how
the encounter is proceeding, and will probably think that the native speaker is
being uncooperative in not giving him the help he needs.20
It would seem that the negative outcome frequently characterizing such
“situations of diVerential power and interethnic stigmatization” (Gumperz
1992c:326) is basically due to participants’ diVerent communicative experi-
ence. Knowledge of the contextualization practices essential for the negotia-
tion of shared understandings can only be acquired by actual participation
over time in discourse communities, communities of learning and practice,
particularly in situations where there is active involvement in “goal-oriented
endeavours” (Gumperz 1996a).21 Where this communicative competence is in

some way lacking, as for example in the case of the South Asian job candidate,
the range of contextualization strategies in the individual’s control is limited,
as is his skill in adapting these strategies to the situation at hand. In social
situations where there are “institutional contacts with majority speakers”
(Gumperz 1992c:327), these diVerences in communicative knowledge may
become valorized and ideologized and in turn reinforce present or develop
into future stereotypes. The stigmatization of South Asian speakers’ commu-
nicative practices thus becomes highly likely and cases of actual discrimi-
nation in decision-making situations will probably occur, in this
way creating a real obstacle to “minority individuals’ success in the society at
large” (ibid.).22
3.2 Rape trial cross-examination
The second case study concerns a rape trial cross-examination. The purpose
of cross-examining, at least in Anglo/American court proceedings, is to re-
veal, through questioning, inconsistencies in the witness’ testimony with re-
gard to topics covered in the direct examination. In other words, the aim is to
‘deconstruct’ the direct examination, and this in fact is what the following
extract deals with.
The two main roles in this particular cross-examination are that of the
defence counsel (C), who is intent on conveying (even though indirectly) the
fact that the victim was actually a consenting participant in the rape attack,
and that of the accuser (W), who claims that she was the unsuspecting victim
of unprovoked aggression. The verbal interaction between these two oppo-
nents can only be fully understood, however, when analyzed in terms of the
adversary situation seen as a whole. In fact, a cross-examination like this can
be seen as a typical Bahktinian situation: the dialogic exchanges between the
counselor and the victim are directed to moving the interaction forward in the
traditional conversation analytic way of turn-by-turn sequences, but, at the
same time, they have the function of persuading the judge and jury (in
Bahktin’s terms, the “supersubject”) to consider their opposite points of view.
According to Gumperz, if the transcript of this cross-examination testi-
mony is studied purely in terms of turn-by-turn sequential analysis, “just
concentrating on regularities of sequential organization across speech ex-
changes” (Gumperz 1992a:231), an incomplete picture of the interactive situ-
ation will be drawn. Even though sequentiality or positioning is obviously an

essential part of any situated interpretation, it represents only one of the
various indexical procedures that inXuence conversational inference. When
considered on its own, and “only overtly lexicalised propositional content” is
used as data (Gumperz 1999b:458), the study of indirect inferencing will be
problematic; for these reasons, Gumperz believes sequential analysis to be
insuYcient as an analytic tool.
To illustrate this point, some examples have been taken from the ex-
changes in the cross-examination in which the defence counsel is questioning
the victim. 23 When asked whether she had had a “fairly lengthy conversation
with the defendant,” she replied, “we:ll we were *all talking;” when confronted
with “well *you knew at that ti:me that the defendant was *in:terested in *you,
didn’t you?”, she answered, “he: asked me how I’(d) *been;” to the question,
“you went to a *ba:r?”, she said, “it’s a *clu:b.” The cross-examination contin-
ued along these lines, with the victim never directly assenting to or denying
the counselor’s insinuations, but instead in some way modifying the way they
had been worded.
If analyzed in conversation analytic terms, the victim’s contributions here
could be considered as instances of other repair (SchegloV, JeVerson and Sacks
1977), W’s substitution of C’s words with her own being seen as a kind of
indirect or “oV-record” correction (Brown and Levinson 1978) of what he was
suggesting, rather than a direct answer to his yes/no questions. From this
analysis, it could be concluded that W was being rather uncooperative, “and in
fact could be accused of interfering with (the defence counsel’s) eVorts to
produce a coherent account” (Gumperz 1995:107). However, this possible act
of repair on the part of the victim does not occur only once or twice, but,
rather strangely for a potentially face-threatening act of this kind, recurs
systematically throughout the whole encounter as a regular sequence.
For Gumperz, it is the “cumulative eVect” (Gumperz 1995:108) of the
victim’s replies that suggests that it is more than just a sequence of repairs,
indicating that other information must be brought into the analysis in order
for the verbal interaction to be fully understood. Even from a purely sequen-
tial analysis perspective, it is necessary to introduce also our knowledge of
what constitutes a cross-examination, and, more speciWcally, of what consti-
tutes a rape trial cross-examination, that is, the “historically and culturally
based knowledge that can be brought to bear on the interpretation of constitu-
ent messages” (Gumperz 1992c:306).
Returning to the testimony transcription, it can be seen how a series of

discursive oppositions of mainly prosodic, but also syntactic nature, are set up
by means of the actual interaction. For example, through the adjacency of
“having a fairly lengthy conversation” and “we were all talking,” a syntactic
opposition is created between the counsel’s suggestion of an intentional act,
“you had a conversation,” and the victim’s description of a state of aVairs, “we
were all talking,” thus conveying the idea of contradiction. If, however, we
limit our analysis to looking at exchanges of this kind in isolation, the full
import of the inferential process does not emerge; it is only through consider-
ing a whole series of such oppositions, in which W systematically selects
certain key aspects of C’s questions and, echoing them, re-presents them in
her own words, that we can begin to infer (using our knowledge of what a
cross-examination involves) that the witness is “intentionally and consistently
opposing her own interpretation to the counsel’s” (Gumperz 1995:109). She
thus appears to be less uncooperative than if she were simply repairing the
counselor’s individual utterances, her ‘narrowing’ or focussing of his ques-
tions in some way negating what he is suggesting, but, at the same time,
through the interaction, conveying a coherent picture of victimhood as op-
posed to that of a consenting participant.
This inferential process is further supported by another kind of discursive
opposition, this time of a prosodic nature. In fact, the particular aspects of the
attorney’s questions that the victim picks out to reformulate are also rhythmi-
cally related, being spaced in such a way that they are almost isochronous. By
isochrony, Gumperz refers here to the “repetitive regular spacing of accented
rhythmic beats across more than two turns of speaking” (Gumperz
1996c:386).24 In isochronous dialogue, the copying or echoing of a previous
speaker’s prosodic contour across turns of speaking, at regular intervals,
indicates agreement with what is being said (Selting 1991, cited in Gumperz
1995:109; cf. Selting 1995) and reinforces the point being made. By actually
changing the rhythmic pattern of the interlocutor’s speech, on the other hand,
the idea of disagreement is given.
During the rape trial cross-examination, the victim and defence counsel’s
diVerent lines of argumentation are reXected and contextualized in part
through the individual rhythmic patterns of their speech. In addition, the
accuser/victim adopts the indirect strategy of rhythmic isochrony to copy and
highlight those parts of the attorney’s speech she wants to work on. For
example, she contrasts “you had some fairly lengthy *conversation” with “we
were *all talking,” and, even more clearly, “you went to a *bar? in *Boston”
with “it’s a *club.” In each of these cases, there is a falling prosodic contour
with the accent being placed towards the end of the utterance. By instinctively
copying this contour in such a way that she is isochronous with these and
other key elements of the counsel’s speech, the victim is able to incorporate
them unobtrusively into her own argument and metaphorically construct her
own version of the story. In so doing, she seems to manage “to neutralize the
counsel’s implicatures and to convey her own perspective on the facts without
appearing overly uncooperative” (Gumperz 1995:109).
3.2.1 Summary
To summarize, throughout this extract of a cross-examination testimony, the
impression of victimhood is conveyed partly by linguistic means, but to a
much greater extent indexically, by means of a process of indirect inferencing.
It is only when this use of indexicality is considered within the context, in
terms of the social factors that govern the conduct of this particular situation,
that it takes on its real signiWcance; in fact, only in these conditions is it
defensible as an essential part of the process of conversational inference.
4. Second language teaching and learning
As a concluding note to this paper, I would like to refer to a speciWc research

area which has been very much inXuenced, both directly and indirectly, by
Gumperz’s work in interactional sociolinguistics: second language teaching
and learning.
It is now well-established that an individual’s knowledge of how to com-
municate, his/her ability to negotiate, to cooperate, is not simply a matter of
producing and recognizing the appropriate grammar and lexis of the target
language. The communicative, or rather, interactional competence which is
“co-constructed” (Jacoby and Ochs 1995) by participants in any conversa-
tional interaction or interactive practice entails much more than this: to
achieve mutual understanding and communicate eYciently in the particular
speech event they are involved in, individuals must also have knowledge of the
interactive and rhetorical strategies through which factual information is
transmitted from speaker to hearer(s) and vice versa (Gumperz and Roberts
1991).25 The level of competence in the use and interpretation of these strate-
gies, particularly with regard to “rhetorical eVectiveness” (Gumperz 1994),

will depend a great deal on the extent to which interactants’ “‘brought along’
context made up of ideological and metapragmatic assumptions” (Roberts
and Sarangi 1999:390) coincide. A crucial part of this “common ground”
(Gumperz 2002) is culture-speciWc, acquired through an individual’s direct
participation in discursive encounters with, for example, members of his/her
family or peer group, as part of the experience of socialization in the commu-
nity he/she is a member of (Gumperz 1997).26
The situation of today’s society, however, is that of a world of rapidly
expanding networks of international contacts, interconnecting increasingly
complex communities from the point of view of people’s cultural origins and
experiences; all this has contributed to a proliferation in issues regarding
intercultural communication. As illustrated in the Wrst case study discussed in
this paper, “in culturally diverse communicative environments, hidden, nor-
mally unnoticed diVerences in perspectives may bring about radically con-
Xicting interpretations of what is happening” (Gumperz 1994:xiv), often
leading to serious intercultural misunderstandings and possible reinforce-
ment of negative stereotypes. This is particularly the case in situations of ‘high
stakes’, “where the interactions are likely to be stressful and protracted and
where the outcomes are critical in people’s lives” (Roberts et al. 1992:89).
Carrying out detailed sociolinguistic analyses of speech events in which
there are frequent contacts between diVerent cultures and between diVerent
ethnic groups is therefore becoming more and more important.27 Gumperz
believes that the results of such research should be made available to all those
involved in second language teaching, so that speciWc cultural content can be
included in language curricula and actual teaching will not be “divorced from
intrasocietal issues of linguistic diversity” (Gumperz 1996b:469). By high-
lighting the ways in which people of diVerent cultural backgrounds interpret
what they see and hear, development of awareness of this kind can become a
great asset for both teachers and learners of a second language. On the one
hand, teachers will beneWt from research-based input on such communicative
conventions in that they will be better equipped to “incorporat[e] that under-
standing into the goals, curriculum design, lessons, and everyday practices of
their classrooms” (SchiVrin 1996:325). On the other, awareness of cultural
diVerences in communication, together with knowledge of interactive strate-
gies and other fundamental aspects of language such as grammar and lexis,
will facilitate students’ interaction with others in everyday life and prepare
them to meet “the cognitive demands of multicultural environments” (Gum-

perz 1996b:480).
Notes
1. This paper was originally inspired by a talk that John Gumperz gave at the 1995 IADA
Round Table on Dialogue Analysis held in Bologna, Italy. It has subsequently been revised
to present an up-to-date picture of Gumperz’s recent work in Interactional Sociolin-
guistics.
2. “Naturally organized activities” are typically face-to-face encounters possessing some
structure recognizable in everyday life (for example, casual chat, job interviews, court
interrogations etc.).
3. Gumperz refers to these mechanisms and patterns of conversational management as
“principles of conversing” (Gumperz 1992c:304).
4. Gumperz, Aulakh and Kaltman 1982, Gumperz, Kaltman and O’Connor 1984, cited in
Gumperz 1992a:232–233.
5. Cf. Tsui’s work on three-part exchanges (Tsui 1989, 1994).
6. Descriptions of this approach can be found in: Silverstein 1992; Mertz and Parmentier
1985; Lucy 1993.
7. For a slightly diVerent and more detailed version of Gumperz’s analysis of these data,
see Gumperz 1992c, 1995.
8. For the construct “typiWcation”, see Schutz 1971, cited in Gumperz 1982:22.
9. For a deWnition of cultural knowledge, see Clark 1992; for a discussion of the diVerent
types of knowledge constituting cultural knowledge, see Gumperz 1995:105.
10. For fuller details on Gumperz’s interpretive level of analysis, see Gumperz and Berenz
1993:92–94.
11. For the description of a case study based on this particular participant observation
experience, see Gumperz 1982.
12. Interactional texts are described by Silverstein as “transcripts that account for all the
communicatively signiWcant verbal and non-verbal signs perceived” (Silverstein 1992, in
Gumperz 2001). For a full description of the transcribing of conversational exchanges, see
Gumperz and Berenz 1993.
13. For a more detailed analysis of this job interview extract, see Gumperz 1995:110–114.
14. Accenting or stress placement is thought to be “crucial in signalling coreference and
alerting interlocutors to what is expected by way of an answer” (Gumperz 1992c:323).
15. Cf. Gumperz’s (2001) analysis of a job interview with an electrician; also in this case,
the interviewers’ indirect accenting strategies are not fully understood by the South Asian
candidate.
16. For a detailed comparison of Indian and Western English prosody, see Gumperz
1982:119–129.
17. For the description of a telephone number test devised to ascertain the occurrence of
these diVerences in stress patterns and the extent to which they constitute a natural
characteristic of South Asian English, see Gumperz 1982:122–123.
18. In some cases, failure by non-native speakers to recognize and/or produce English
stress patterns can also be grammatically motivated, as, for example, in speakers of Hindi,
who convey the idea of emphasis in part by the discourse particle ‘hi’ (Gumperz, Aulakh
and Kaltman 1982; see also discussion in Roberts et al. 1992:92).
19. Cf. Tyler’s (1995) study of miscommunication between a Korean tutor and a student
who is a native speaker of American English. In this case, in order not to appear rude, it is
the Korean tutor who plays down his knowledge of a particular skill.
20. It is important to note, however, that these ‘criticisms’ are not directly attributable to
participants’ personal feelings or beliefs, but are instead “situated reactions” provoked by
“indexical ties between verbal form and context that function metapragmatically to create
interpretive eVects” (Gumperz 1996c:400–401).
21. In his talk on the concept of communicative competence in culturally complex societ-
ies, Gumperz proposed a deWnition of communicative competence as “the ability to create
and maintain communicative involvement” (Gumperz 1996a).
22. Cf. Gumperz 1999a for the construction of a cultural defence in a court case involving
a Northern California Indian.
23. Transcription conventions are taken from Drew 1992. Gumperz has added ‘*’ to
indicate syllable stress.
24. Gumperz is following the deWnition of isochrony advocated in studies on prosody as
an interactive phenomenon (Couper-Kuhlen and Auer 1991, Uhmann 1996).
25. For a comparison of the concepts of communicative and interactional competence,
see He and Young 1998.
“Interactive practices” are deWned as “structured moments of face-to-face interaction [...]
whereby individuals come together to create, articulate, and manage their collective
histories via the use of sociohistorically deWned and valued resources” (Hall 1995:207–
208).
26. For example, see Gumperz’s analysis of shared inferencing between two sisters in an
informal coVee shop conversation (Gumperz 2000 and 2002).
27. For example, see the studies Gumperz has carried out with others in bilingual, “coop-
erative learning” classrooms to investigate the peer group inferential processes students
rely on to establish collaborative interaction with each other (Gumperz and Field 1995;
Gumperz, Cook-Gumperz and Szymanski 1999).
References
Brown, Penelope and Levinson, Stephen C. 1978. “Universals in language usage: Polite-
ness phenomena”. In Questions and Politeness: Strategies in Social Interaction, E.N.
Goody (ed.), 56–311. Cambridge: Cambridge University Press.
Clark, Herbert H. 1992. Arenas of Language Use. Chicago: University of Chicago Press.
Couper-Kuhlen, Elizabeth and Auer, Peter. 1991. “On the contextualizing function of
speech rhythm in conversation”. In Levels of Linguistic Adaptation, J. Verschueren
(ed.), 1–18. Amsterdam and Philadelphia: John Benjamins.
Drew, Paul. 1992. “Contested evidence in courtroom cross-examination: The case of a
trial for rape”. In Talk at Work: Interaction in Institutional Settings, P. Drew and J.
Heritage (eds.), 470–520. New York: Cambridge University Press.
GoVman, Erving. 1976. “Replies and responses”. Language in Society 5:257–313.
Goodwin, Charles and Heritage, John. 1990. “Conversation analysis”. Annual Review of
Anthropology 19:283–307.
Gumperz, John J. 1990. “Conversational cooperation in social perspective”. Parasession
on the Legacy of Grice. Berkeley Linguistic Society 16:429–441.
Gumperz, John J. 1992a. “Contextualization and understanding”. In Rethinking Context:
Gumperz, John J. 1992b. “Contextualization revisited”. In The Contextualization of Lan-
Benjamins.
Gumperz, John J. 1992c. “Interviewing in intercultural situations”. In Talk at Work:
Interaction in Institutional Settings, P. Drew and J. Heritage (eds.), 302–327. New
York: Cambridge University Press.
Gumperz, John J. 1993. “Culture and conversational inference”. In The Role of Theory in
Language Description, W.A. Foley (ed.), 193–214. Berlin and New York: Mouton de
Gruyter.
Gumperz, John J. 1994. “Foreword”. In Crosstalk and Culture in Sino-American Commu-
nication, L.W.L. Young, xiii-xix. Cambridge: Cambridge University Press.
Gumperz, John J. 1995. “Mutual inferencing in conversation”. In Mutualities in Dialogue,
I. Marková et al. (eds.), 101–123. Cambridge: Cambridge University Press.
Gumperz, John J. 1996a. “Communicative competence in culturally complex societies”.
Paper presented at invited colloquium on Creating Communicative Competence at
1996 Conference on Discourse Communities, Chicago: American Association for Ap-
plied Linguistics.
Gumperz, John J. 1996b. “On teaching language in its sociocultural context”. In Social
Interaction, Social Context, and Language. Essays in Honor of Susan Ervin-Tripp, D.I.
Slobin et al. (eds.), 469–480. Hillsdale, NJ: Lawrence Erlbaum.
Gumperz, John J. 1996c. “The linguistic and cultural relativity of conversational infer-
ence”. In Rethinking Linguistic Relativity, J.J. Gumperz and S.C. Levinson (eds.), 374–
406. Cambridge: Cambridge University Press.
Gumperz, John J. 1997. “On the interactional bases of speech community membership”.
In Towards a Social Science of Language. Papers in Honor of William Labov, Vol. II:
Social Interaction and Discourse Studies, G.R. Guy et al. (eds.), 183–203. Amsterdam
and Philadelphia: John Benjamins.
Gumperz, John J. 1999a. “Culture in the cultural defense”. Proceedings of the Sixth Annual
Symposium about Language and Society (1998), SALSA VI:115–132. Austin: Depart-
ment of Linguistics, University of Texas.
Gumperz, John J. 1999b. “On interactional sociolinguistic method”. In Talk, Work and
Institutional Order, S. Sarangi and C. Roberts (eds.), 453–471. Berlin and New York:
Mouton de Gruyter.
Gumperz, John J. 2000. “Inference”. Journal of Linguistic Anthropology 9 (1–2):131–133.
Gumperz, John J. 2001. “Interactional Sociolinguistics: A personal perspective”. In The
Handbook of Discourse Analysis, D. SchiVrin, D. Tannen and H.E. Hamilton (eds.),
215–228. Oxford: Basil Blackwell.
Gumperz, John J. 2002. “Sharing common ground”. In Soziale Welten und Kommuni-
kative Stile: Festschrift für Werner Kallmeyer zum 60. Geburtstag, I. Keim and W.
Schütte (eds.), 47–56. Tübingen: Gunter Narr.
Gumperz, John J. and Field, Margaret. 1995. “Children’s discourse and inferential prac-
tices in cooperative learning”. Discourse Processes 19:133–147.
Gumperz, John J. and Roberts, Celia. 1991. “Understanding in intercultural encounters”.
In The Pragmatics of Intercultural and International Communication, J. Blommaert
and J. Verschueren (eds.), 51–90. Amsterdam and Philadelphia: John Benjamins.
Gumperz, John J. and Toribio, Almeida J. 1999. “Codeswitching”. In The MIT Encyclope-
dia of the Cognitive Sciences, R.A.Wilson and F.C. Keil (eds.), 118–120. Cambridge,
MA: The MIT Press.
Gumperz, John J., Aulakh, Gurinder and Kaltman, Hannah. 1982. “Thematic structure
and progression in discourse”. In Language and Social Identity, J.J. Gumperz (ed.),
22–56. New York: Cambridge University Press.
Gumperz, John J., Cook-Gumperz, Jenny and Szymanski, Margaret. 1999. Collaborative
Practices in Bilingual Cooperative Learning Classrooms. CREDE Research Report 7.
Santa Cruz, CA and Washington DC: Center for Research on Education, Diversity
and Excellence.
Gumperz, John J., Kaltman, Hannah and O’Connor, Mary C. 1984. “Cohesion in spoken
and written discourse: Ethnic style and the transition to literacy”. In Coherence in
Spoken and Written Discourse [Advances in Discourse Processes XII], D. Tannen
(ed.), 3–19. Norwood, NJ: Ablex.
Hall, Joan K. 1995. “(Re)creating our worlds with words: A sociohistorical perspective of
face-to-face interaction”. Applied Linguistics 16:206–232.
He, Agnes W. and Young, Richard. 1998. “Language proWciency interviews: A discourse
approach”. In Talking and Testing: Discourse Approaches to the Assessment of Oral
ProWciency, R. Young and A.W. He (eds.), 1–24. Amsterdam and Philadelphia: John
Benjamins.
Jacoby, Sally and Ochs, Elinor. 1995. “Co-construction: An introduction”. Research on
Language and Social Interaction 28 (3):171–183.
Lucy, John A. (ed.). 1993. ReXexive Language: Reported Speech and Metapragmatics.
Mertz, Elizabeth and Parmentier, Richard J. 1985. Semiotic Mediation. Orlando: Aca-
demic Press.
volume.
Roberts, Celia and Sarangi, Srikant. 1999. “Introduction: Revisiting diVerent analytic
frameworks”. In Talk, Work and Institutional Order, S. Sarangi and C. Roberts (eds.),
389–400. Berlin and New York: Mouton de Gruyter.
Roberts, Celia, Davies, Evelyn and Jupp, Tom. 1992. Language and Discrimination: A
Study of Communication in Multi-Ethnic Workplaces. London: Longman.
SchegloV, Emanuel A., JeVerson, Gail and Sacks, Harvey. 1977. “The preference for self-
correction in the organization of repair in conversation”. Language 53:361–382.
SchiVrin, Deborah. 1994. Approaches to Discourse. Oxford: Basil Blackwell.
SchiVrin, Deborah. 1996. “Interactional sociolinguistics”. In Sociolinguistics and Language
Teaching, S.L. McKay and N.H. Hornberger (eds.), 307–328. Cambridge: Cambridge
University Press.
Schutz, Alfred. 1971. Collected Papers II: Studies in Social Theory. 3rd printing. The
Hague: NijhoV.
Selting, Margret. 1991. Prosodie im Gespräch. Habilitationsschrift, Universität Oldenburg.
Selting, Margret. 1995. Prosodie im Gespräch. Aspekte einer Interaktionalen Phonologie der
Konversation. Tübingen: Niemeyer.
enough?” In The Contextualization of Language, P. Auer and A. Di Luzio (eds.), 55–
76. Amsterdam and Philadelphia: John Benjamins.
Tsui, Amy B.M. 1989. “Beyond the adjacency pair”. Language in Society 18 (4):545–564.
Tsui, Amy B.M. 1994. English Conversation. Oxford: Oxford University Press.
Tyler, Andrea. 1995. “The coconstruction of cross-cultural miscommunication. ConXicts
in perception, negotiation, and enactment of participant role and status”. Studies in
Second Language Acquisition 17(2):129–152.
Uhmann, Susanne. 1996. “On rhythm in everyday German conversation: Beat clashes in
assessment utterances”. In Prosody in Conversation: Interactional Studies, E. Couper-
Kuhlen and M. Selting (eds.), 303–365. Cambridge: Cambridge University Press.
Response essay 105
Chapter 8
Response essay
John J. Gumperz
Let me begin with a little more background on the issues and Wndings that Wrst
motivated my research on interaction. As Levinson (this volume: 31) puts it in
discussing my approach to contextualization, “Readers will not understand
his work if they view it just as the study of conversation” or, for that matter,
“interaction”, I would add. When I replied to Prevignano’s initial question,
stating that my interests are both theoretical and applied, I meant to suggest
that I did not set out to study interaction in the abstract. What perhaps I did
not make clear was how my current theoretical thinking about language and
communication builds directly on earlier ethnographic Weld studies that led
me to conclude that I needed to learn more about interaction in order to deal
with the questions I had set out to resolve.
I have always had a strong interest in linguistic and cultural diversity, and
in particular in questions such as: How are sociocultural boundaries repre-
sented in language? How does this relationship aVect human understanding
and especially people’s ability to convey information, persuade others and
carry on their everyday aVairs? What is its import for our understanding of
how social institutions work?
Before responding to speciWc comments, I will brieXy describe, or, more
appropriately perhaps, reconstruct, how such interests led me to concentrate
on a theory of interaction. When I Wrst embarked on Weld research, the
nineteenth century assumption that human populations come naturally di-
vided into distinct, bounded groups was still widely accepted. It was assumed
that each of such units can be treated as an island unto itself, with its own
language or dialect and culture, which must be analyzed in their own terms.
Communication across group boundaries was said to be problematic, and
linguistic diversity was seen as potentially a major obstacle to societal consen-
sus. Yet, apart from abstract discussions about translation as the process of
converting the morpho-syntactic or semantic structures of one language into

another, little in the way of research had been done to clarify what linguistic
and cultural diversity involved at the level of everyday communication. My
aim was to collect veriWable data on both linguistic and socio-cultural repre-
sentations of diversity.
I was familiar with the research of linguistic geographers in Europe and
the United States who had convincingly demonstrated that current linguistic
boundaries had arisen in the course of the preceding centuries as a function
of the shifts in power relations, as well as of the related political, economic
and religious forces which together led to the development of modern nation
states (BloomWeld 1933). The new urban formations produced by these
changes in eVect served to concentrate previously dispersed sources of
inXuence into centralized, politically, economically and culturally powerful
centers of innovation, which then radiated pressure towards uniformity into
the surrounding hinterland. In this way, ecological conditions arose that led
to the redirection of interpersonal relations inwards, towards the dominant
center. The resulting communicative environments brought about a reduc-
tion of local diversity. Nationalist and religious ideologies, along with their
associated new national language varieties, circulated within these zones of
inXuence, while, at the same time, boundaries between separate zones were
sharpened, further accelerating the homogenization process. In other words,
present-day linguistic boundaries cannot be naturalized and accepted as pre-
existing facts. They basically reXect the human history of the past few centu-
ries and the forces that led to the current situation, which, in turn, we
must study.
Reviewing these developments in his classic articles on the speech com-
munity, Leonard BloomWeld (1933) argued that the degree of linguistic diver-
sity in any one region is a function of the patterning of interpersonal contacts
over time. His formulation suggests an initial outline of a theory of diversity
that rests on human interaction as an analytical prime, and does not rely on a
priori assumptions about ethnic, class or group identity. Such a theory, among
other things, provides explanations for how language distinctions like those
between Dutch and German, French and Italian, Serbian and Croatian, and
many others throughout the world, came about in the course of history, as
a result of constraints on human contact imposed by changing power rela-
tionships. To be sure, the linguistic geographers’ research on which these
arguments were based had been conWned to populations speaking histo-
Response essay 107
rically-related language varieties. But since the theory focusses on interper-

sonal relationships and their long-term historical eVects, not on how they are
represented in speciWc languages, there seemed to be no reason why similar
reasoning should not be extended to multilingual situations, provided appro-
priate social conditions existed.
In my early Weldwork, I set out to explore the validity of the interactional
approach to diversity, Wrst in a caste-stratiWed North Indian village and then
in a small, culturally homogeneous, Northern Norwegian town with a strong
egalitarian ideology (Gumperz 1971). In the North Indian village, for ex-
ample, neither the divisions among named castes, nor the politically salient
religious distinctions between Hindus and Muslims were directly reXected in
linguistic signs. I did, however, discover signiWcant bundles of linguistic
isoglosses. One set of features divided the majority of local touchable agricul-
turist and artisan castes who shared a single speech variety from the three
untouchable castes. Each of the latter, untouchable groups in turn was set oV
by additional sets of isoglosses from the other two. When I re-examined the
linguistic data in the light of ethnographic information on interpersonal
relations, it became apparent that touchable village residents formed a single
communicative whole, where intercaste contacts were relatively frequent and
children freely entered into mixed-caste neighborhood playgroups. Untouch-
ables, on the other hand, lived in separate neighborhoods and did not main-
tain informal peer relations with other castes. Based on such Wndings, I
hypothesized that interaction is most probably the determining factor in the
construction and maintenance of the intra-village linguistic distinctions. As
with the linguistic geographers’ isoglosses, the relevant signs were grammati-
cally not very signiWcant. Nevertheless, they carried great social import.
However, contrary to what my reading of the literature had led me to
expect, frequency of interactive contacts alone could not explain the facts on
the ground. Untouchable caste members tended to spend most of their days
working in the households of their touchable employers and had only limited
opportunities for socializing with peers. Yet, they did not adopt their more
prestigious employers’ speaking practices, nor for that matter did they give up
their own distinct manners and dress. To explain this, I argued that, apart
from density of contact, ideologies of interpersonal relations and normative
principles constraining what could be said to whom and how it should be said
must also be at work. In other words, at issue is not just the number of times
individuals speak to others. The quality of the interaction, that is to say, what
is said, when, and by whom in the course of such encounters, seemed to count
as an equally important factor.
The Norwegian study yielded additional conWrmation for the interaction
hypothesis (Blom and Gumperz 1972). The ideology of egalitarianism here
was so strong that residents refused to acknowledge the existence of sig-
niWcant class diVerences in the locality. And indeed it soon became evident
that income and occupational status distinctions were not clearly reXected in
talk. Nevertheless, it was possible to detect signiWcant diVerences, this time
not at the level of phonology or syntax, but in patterns of language use and
particularly in code-switching between the local dialect and the regionally
dominant variety of Standard Norwegian. The diVerences in language use
distinguished those who, by virtue of their occupation, maintained regular
and close relationships with speakers in the surrounding small towns and in
the more highly urbanized South of the country, from other residents whose
signiWcant contacts were conWned largely to local ones. Once more, density
and quality of interpersonal contacts turned out to be more signiWcant than
social categorizations.
Furthermore, as Wndings from research in previously little-known lin-
guistic areas became better understood, it also became evident that interac-
tion-based forces of linguistic convergence and diVerentiation do not
necessarily respect language boundaries. Throughout the known world, both
in large nation states, as well as in less urbanized regions, where small, sepa-
rate, localized language/culture units live among majority populations, regu-
lar and frequent translocal, interpersonal contacts set oV pressure for change.
Over time, these pressures tend to blur and ultimately eliminate established
lexico-grammatical distinctions (Gumperz 1996; Gomez-Imbert 1996).
A third ethnographic study of a multilingual village at the Indo-Aryan/
Dravidian language border largely supports this interpretation. In this Central
Indian site, the dominant group of landowners spoke a dialect of Kannada, a
Dravidian language, while Indo-Aryan Marathi was the oYcial administrative
language in the surrounding region and the medium of schooling. A sig-
niWcant minority population of artisan and untouchable castes, moreover,
spoke dialects of Urdu (Indo-Aryan) or Telugu (Dravidian). We were struck
by the frequent code-switching observed in everyday village talk. For example,
just as the small town residents in Norway had alternately used Standard
Norwegian and local dialect forms, people in this multilingual village readily
switched between the historically-unrelated Marathi or Urdu and Kannada,
Response essay 109
often in the course of a single brief encounter. And they did so for similar
communicative ends.
When we compared transcripts of everyday talk, moreover, we found that
texts in the local dialects of Urdu and Kannada could be described by a single
underlying set of surface syntactic rules. What local peoples referred to as two
separate languages were analytically distinct, for the most part, only at the
level of word morphology and lexicon, so that translation from one local
variety to the other was largely a matter of word-for-word substitution. In
other words, for the speakers involved in these situations, code-switching was
grammatically not more onerous than dialect-switching in the Norwegian
case. Apparently, centuries of co-residence in the same locality had brought
about large-scale structural convergence among local speech varieties. Yet, the
changes had stopped short of complete creolization. If comprehensibility is
not an issue, how is it that such structurally homologous varieties maintain
themselves in this relatively small population in the face of long-term close
contact? As the code-switching evidence shows, it is not because people
cannot understand each other. We are dealing with a socially-motivated lin-
guistic ideology of diVerentiation, a set of local norms similar to those found
in the North Indian situation, which keep speakers from taking over the
others’ modes of speaking.
The picture of variability that emerges from these Weld studies is quite
diVerent from the conventional one. Common practices that accept taken-
for-granted notions of community, explaining language usage in relation to a
presumably uniform, community-wide system of beliefs and values, that is,
treating each as a self-contained island, are seriously in need of revision.
Whereas truly isolated, homogenous communities have by now all but disap-
peared, as the pace of population migration continues to grow, multilingual or
multidialectal communicative economies like those I have sketched above are
rapidly becoming the rule rather than the exception. It would be more fruitful
therefore to turn to a new base metaphor for sociolinguistic research, a meta-
phor that incorporates diversity into the research design. One might for
example focus on a set of local formations marked by translocal economic,
political, religious and other similar networks of relationships. All this is not
to claim that communities do not have a basis in linguistic reality. I am
arguing that communities must be seen as ideological constructs. Whereas
communicative practices rely on common ground, that is, shared assump-
tions to achieve understanding, ideologies are by their very nature contested.
Communities on one deWnition are treated as the oYcial units for much
governmental and educational planning, but how these communities are
deWned is always open to question, and cannot be taken for granted. Ulti-
mately, communicative practices that are part of a daily reality may have an
inXuence on governmental decisions. I am arguing that the relationships
between ideological constructs and practices are the most useful point of entry
for an investigation of how societal level forces can aVect and be aVected by
everyday talk.
In my earlier work, I used the term “linguistic repertoire” to refer to the
speaking practices of social collectivities, arguing, as di Luzio (this volume: 2–
3) points out, that repertoires are describable in terms of sets of functionally-
diversiWed speech varieties. I found the notion of repertoire useful because it
served to call attention both to diVerences among constituent varieties and to
commonalities due to convergence. To show, in other words, that we are
dealing with what, for want of a better term, we can call “communicative
economies”, that have their own local and translocal forms of social and
linguistic organization. Recent, more detailed work on everyday communica-
tion, however, has led me to doubt that it is useful to assume that discourse
level diversity can be explained in terms of alternation among sharply-
bounded speech varieties. I will return to this point later.
The ethnographic insights that initially seemed most in need of explana-
tion were the following. First, the surprising lack of correspondence between
people’s perceptions of boundaries and the linguistic nature of the boundary
markers. In none of the three Weld situations did we Wnd a direct one-to-one
relationship between what I originally termed “socially”, but would now call
“ideologically” grounded language/dialect distinctions, on the one hand, and
units of language, deWned by grammatical distance alone, on the other. Both
in multilingual and in multidialectal settings, what from a purely linguistic
perspective may count as minor distinctions can often, for largely ideological
reasons, attain great social import as badges of identity. If structure alone
cannot explain such communicative facts, how do we look at verbal interac-
tion to Wnd explanations? Secondly, people readily communicated across
linguistic boundaries. What locals refer to as language or dialect distinctions
do not by themselves prevent understanding, or as such limit interaction. In
the early sites, as well as in later European and U.S. Weld work, local speech
varieties formed part of a single communicative economy, where, depending
on circumstances, switching among distinct languages, dialects or even styles
Response essay 111
of the same language can produce communicatively equivalent eVects. We are

dealing with socially-motivated ideologies of diVerentiation, localized norms
akin to those in the North Indian situation, that in some cases lead to conver-
gence, and in other conditions keep speakers from taking over their interlocu-
tors’ speaking habits.
Linguistic and cultural boundaries are not just ‘naturally’ there, they are
communicatively and, therefore, socially constructed. Thus, they cannot be
essentialized and treated as self-contained islands in research on communica-
tive practices. Apart from interaction as such, ideology, power and history are
all central to the way diversity works; depending on how these factors interre-
late in speciWc circumstances, interaction can serve either to accentuate or
attenuate the eVects of diversity.
I therefore decided that I needed to learn more about how and through
what kind of linguistic knowledge people understand each other in everyday
talk, as well as how this knowledge is acquired and how it circulates, before
attempting to go on to further generalizations. Ethnographers of communica-
tion were the Wrst to turn away from community-based sociolinguistic analy-
sis. They argued that, while grammatical analysis has advanced greatly in
depth and sophistication since the nineteen sixties, we have little if any com-
parative information about everyday communication (Gumperz and Hymes
1972). The problem was not just that more of the world’s many languages and
cultures were in need of detailed study. Rather, our sociolinguistic theories
were unable to deal with many signiWcant facts of language use in today’s
inherently diverse societies. It seemed important, therefore, to bypass estab-
lished structural-functionalist research designs and Wnd ways of developing
integrated semiotic approaches that would focus directly on speciWc events as
the sites where language and culture meet.
In my work on interaction, I built on my experience with ethnography of
communication, relying on in-depth participant observation and interview-
ing to discover a range of naturally-occurring communicative situations or
events. I concentrated on events that typically involved talk on issues of
general interest, in settings where either my collaborators or I were suYciently
familiar with the relevant cultural and historical facts. As I see it, ethnography
is not simply a matter of investigators creating written accounts of their
experiences; it is basically a reXexive hypothesis formation, hypothesis testing
and reformulation process. Building on background knowledge acquired in
this way, I embarked on comparative case study analyses, contrasting in-
stances of “intercultural communication” with other situations where partici-

pants were able to rely on shared communicative experience to determine the
relevant linguistic signs. The procedure is akin to those employed in linguistic
Weld work, where grammatical and ungrammatical sentences are compared to
gain understanding of grammatical processes.
Initial insights into the semiotic/cognitive aspects of diversity came from
code-switching. Both in the Norwegian and the Central Indian village, many
speakers typically spoke both a local variety and a standard variety, alternating
between the two in response to the communicative demands of the exchange
in which they were engaged. In the Norwegian case, where we Wrst looked at
the phenomenon in some detail, we discovered two distinct patterns. In the
Wrst, situational switching, code alternation, co-occurs with or signals shifts in
local, participant alignments or topic change (Blom and Gumperz 1972;
Gumperz 1982). But there is a second pattern, where words or phrases of a
second language are inserted into what at the level of discourse must count as
the same conversational string. Using as a basis an analysis of preceding and
following talk, I hypothesized that the juxtaposition here counts as a momen-
tary allusion to, or evocation of, a diVerent set of contextual presuppositions,
which has communicative import and aVects the listener’s interpretation. I
initially used the term “metaphoric switching” to suggest that the communi-
cative eVect of these switches is similar to metaphors in their rhetorical eVect.
A code-switch of the latter kind can then be seen as a communicative
strategy, an attempt to inXuence the listener’s interpretation of what the
speaker intends to convey. My search for similar discursive phenomena in a
range of situations, and attempts to explain their semiotic functioning, led to
the approach to conversational inference I outlined in the discussion with
Prevignano and di Luzio. The notion of contextualization cues was developed
in the course of the work on intercultural communication as a way of general-
izing insights gained through code-switching analysis (Gumperz 1982). Note
that, as Levinson (this volume: 36) points out, contextualization cues consti-
tute a natural class of signs that cuts across generally recognized channels of
linguistic signaling. They function as co-occurring clusters of signs that aVect
interpretation by evoking the contextual presuppositions in terms of which
meaning is assessed. This suggests that we need to deal with understanding at
a level of discourse organization which is distinct from more commonly
studied lexico-semantic signaling processes. My use of the term “discursive
practice” is akin to Hanks’ use of the term in his Language and Communicative
Response essay 113
Practices (1996), which presents detailed arguments to support the need for
the above distinction.
My approach to semiotic phenomena is broadly in line with Michael
Silverstein’s Peircian semiotic functionalism. Silverstein argues, among other
things, that contextualization cues are a class of indexicals (1992); that is to
say, they function via vector-like cognitive processes radiating out in certain
directions from a known origin, but have no deWnite terminus (Silverstein
1993). Yet, in spite of its inherent indeterminacy, indexicality, as most of
today’s linguistic anthropologists will agree, is crucial for anyone concerned
with discursive representations of culture. Among other things, indexicality
suggests an explanation for what I referred to above as the lack of Wt between
grammatical distance and ideologically-based language distinctions. What
indexes do is act as Xags, cues or reminders to listeners to search their memory
for possible alternative ways of explaining or framing what they hear or
otherwise perceive or recall. That is to say, to arrive at explanations that make
coherent sense of what is going on in the situation at hand. There is thus no
generally agreed-upon, stable meaning relationship between the indexical
sign and any speciWc explanation. The interpretations are always highly con-
text-speciWc in that they depend on aspects of the communicative process that
resist strict formalization.
A key aspect of Peircian theory, not suYciently foregrounded in the
currently available literature, is that signs have meaning only by virtue of
being taken to stand for an object by some interpreter (Lucy 1993). This lays
the groundwork for an integrated semiotic framework, capable of dealing with
both the linguistic and socio-cultural aspects of everyday communication
within a single analytical perspective. Saussurean denotational meanings must
ultimately always be seen as what they are, abstracted from situated practice in
accordance with procedures that set aside the very signaling processes on
which our assessment of the situated interpretation of communicative intent
rests. Useful as they are for comparative grammatical analyses and historical
reconstruction, denotational meanings cannot by themselves explain every-
day human action, partly because they do not account for discourse-level
signaling processes. Once we look at verbal communication as relying on both
symbolic and indexical signs, we can see that human interpreters become
central to the interpretive process. That is to say, the interpreters’ communi-
cative background, their ability to retrieve background knowledge while en-
gaged in an interaction, the power relations aVecting access to relevant
knowledge, and the ideological social and cultural constraints that aVect
understanding, all become integral constituents of the communicative pro-
cess. Although I have not explicitly said so before, I basically share this
perspective in my attempt to develop ways of dealing with interaction that
account for universal, as well as culturally and situationally-speciWc, aspects
of discourse.
Turning now to the speciWc comments, Levinson, Eerdmans and, to
some extent, Prevignano, all expand on my responses in the discussion.
Levinson explores the notion of contextualization from a more formal, that
is, more linguistically-oriented, pragmatic perspective. His discussion of
“context importation and context invocation” is useful and illuminating in
that it cites a number of examples from the linguistic literature to suggest
that the notion of contextualization cue is not as far removed as one might
think from more familiar linguistic pragmatic processes. One might say that
contextualization applies a pragmatic perspective to discourse analysis. But
recall that, especially in my current work, I focus on events or situations of
speaking where participants are treated as engaging in discursive practices,
rather than speaking particular languages. Participants in an event engage in
communicative practices, where verbal expressions, indexical signs like
code-switching, prosody or non-verbal signs, and other signaling modes
may all achieve similar communicative eVects. It is this notion of discursive
or communicative practice, which, in accord with Hanks (1996), I take to be
distinct from language as such, that is relevant here. It enables us to achieve
the broader, more inclusive level of semiotic analysis, not tied to the speciWcs
of particular linguistic signs, which we need to treat code-switching, non-
verbal signs, dress and other signaling modes as contributing to the interpre-
tation process. Along with Hanks, GoVman, conversational analysts and
others, I think of discursive practice as a separate level of analysis and would
argue that Levinson, illuminating as his points are, takes a more language-
oriented perspective; while, from this perspective, the distinction between
diVerent kinds of contextualization makes sense, it is less important at the
level of communicative practice.
Finally, as to what Levinson (this volume: 38) calls the “ineVability” of
understanding, I agree that my answers to Prevignano’s “persistent” questions
as to how to deal with the ambiguity that arises for the analyst as a result of the
complexities of the interpretive process seemed somewhat evasive. Let me try
to be more speciWc and rephrase Levinson’s second question (this volume: 38)
Response essay 115
as follows: What gives the analyst the right to say, “B perceives that A said this,
but he thinks A intended to convey that”? As pointed out before, I do not
claim that Interactional Sociolinguistic (hereafter, IS) analysis can provide
deWnite answers to such questions. Since situated understanding ultimately
always builds on indexical signs, we cannot avoid ambiguity. But then, as
Prevignano (this volume: 17) points out, even properly socialized natives can
understand diVerent things by the same utterances. In fact, we all know
political discourse, advertising, committee negotiations and similar genres
rely on ambiguities for rhetorical eVect (Cook-Gumperz and Gumperz 1994).
Interpretation, moreover, is not a unitary assessment. Typically, inferences
are interactively negotiated over several turns at speaking and such negotia-
tions rarely yield uncontestable understandings. As I see it, the IS analyst’s
task is to reconstruct the understanding process, relying on a varied array of
formally distinct sources of knowledge both at the level of content and form.
Let me turn once more to examples from my published writings to illus-
trate how I view the interpretive process: (i) in the cross-examination ex-
change described by Eerdmans (this volume:94–97), we understand what is
intended by virtue of the fact that we can check our local inferences against
widely-known background presuppositions about what goes on in a cross-
examination and what we can expect the individual participants to have
intended to achieve by means of their talk; (ii) take the case of the graduate
student from Discourse Strategies which I referred to in the discussion
(Prevignano and di Luzio, this volume:19–20), who responded, “Ahma git
me a gig,” when I asked him to see me in my oYce. I was at Wrst puzzled by
what seemed like an unexpectedly jarring code-switch to Black English, which
did not Wt into my initial understanding of what had transpired. It was only
when I compared my interpretation of the incident with comments of other
black, as well as white, students, that I arrived at a reasonable interpretation.
An interpretation, moreover, that Wts in quite well with what I later learned
about African American communicative practices in my work with racially-
mixed school populations; (iii) in a conversational exchange among American
Asian graduate students (Gumperz 1992b:48–9), A asks B what his brother’s
major in college is. When B answers “ah, bio*chem?”, A replies “ ...... *da::(ng),
.. the *bio, huh?”. And when B conWrms, he goes on with “we Asians man! I’m
telling you.” Although when I Wrst looked at the exchange, I did not quite
understand what the talk was about, it was clear from the way the two partici-
pants responded to each other and jointly went about developing a shared
theme that they understood each other. A few turns later in the exchange, it
appeared that what the group was concerned with was the failure of Asian
professionals to take leadership roles in public aVairs or in other technical
Welds; (4) in the Wnal example (Gumperz 1996:388V), an older South Asian
student (D) came to see a lecturer (L) in an adult education center to ask about
enrolling in a new course in a nearby community college. When the lecturer
told him that, although she had participated in planning the course, she was
not involved with student admission, he directly contradicted her, arguing
that she did have a role in the selection process. He kept on with this for
several minutes, totally contradicting nearly all of her statements, and seem-
ingly oblivious of the fact that she was becoming more and more annoyed.
Taken at its face value, D’s behavior seemed at Wrst quite inexplicable both to
me and to the two native South Asian assistants who had helped with the
transcription. It was only when we stepped back from the details of the
exchange, trying to imagine what someone like D might have been seeking to
accomplish with his repeated denials, asking ourselves why he would want to
oVend someone he was asking help from, that we came up with a likely
solution. The assistants pointed out that D sounded as if he was pleading. And
as they said this, I was reminded of an encounter with an older man in a North
Indian village who asked me to help his son, a college student, obtain an
assistantship in the United States. When I tried to tell him that, as a junior
professor, I did not have the right connections, he replied “you can do every-
thing,” using an expression remarkably similar, i.e. intertextually-related, to
some of the statements D employed.
What is analyzed here are events made up of a set of utterances or speech
exchanges and the surrounding circumstances relevant to their interpretation.
The aim is to arrive at interpretations that a) account for what transpires in the
event as a whole and b) Wt into what we know independently about local
ideologies of language and interpersonal relations. The discussion highlights
the functioning of shared background knowledge or common ground in the
interpretive process. Ethnographic observation is clearly important as the
means by which the analyst acquires the knowledge of background presuppo-
sitions and of the indexical conventions that participants tend to rely on in
deriving what they assume is intended, i.e. Levinson’s ‘that’, from what they
perceive as ‘this’. But what I am referring to here is a type of ethnography of
communication that focusses on what people do in speciWc types of situations,
how they react to others and how they talk about it. Along with other
Response essay 117
interpretivists, I rely on triangulation, comparing my own with others’ inter-

pretations or examining what happens at one point in the exchange to deter-
mine what coheres with or Wts into what went on before or after. Moreover,
whenever possible, I look for intertextuality to suggest possible interpreta-
tions. In contrast to other interpretive traditions that focus only on content,
however, IS analysis then seeks to reWne such possible interpretations in the
light of what is known about the indexical value of the contextualization cues
used, the utterance’s positioning within the stream of talk and the responses it
receives. Interpretation, when seen in the IS perspective, integrates the so-
called formal and content-level aspects of discourse. All such factors together
act as a check against misunderstandings, so that in a large number of cases we
can be fairly conWdent of our interpretations.
I believe that interactional sociolinguistic analysis can provide valuable
insights into the culturally-speciWc rhetorical traditions that underlie our
reasoning, into how understandings come about and into what it is about
interaction that can lead to misunderstandings.
In her contribution, Eerdmans writes from the perspective of an applied
linguist interested in analytical methods. As she points out, interactional
sociolinguistic analysis relies on sequentially-organized conversational ex-
changes like those of conversational analysis, as well as on Gricean notions
like implicature and inference. I should add that this combination of method-
ological approaches does not just constitute eclecticism on my part. My no-
tion of inference informally builds on Grice in that I assume that what is at
issue is “what a speaker intends the listener to do with the message,” so that
communication is always goal-oriented. Furthermore, I believe that while
listeners’ interpretations, as they emerge in discourse, are often quite far
removed from denotational meanings, they can always be derived from what
was literally said.
Basically, I regard myself as a linguistic anthropologist for whom the
study of interaction is integral to his more comprehensive ethnographic in-
vestigations of the often taken-for-granted ways in which local populations
deal with issues they encounter in going about their aVairs. That said, it is true
I share the conversation analysts’ position that talk is constituted by sequen-
tially-organized conversational exchanges, and that conversation in a way
creates its own communicative ecology. But consider my deWnition of conver-
sational inference: “the situated or context-bound process of interpretation,
by means of which participants in an exchange assess [each] others’ intentions
[at any one point in an interaction], and on which they base their responses”
(Gumperz 1982:153). Given this deWnition, we cannot take understanding for
granted and must assume that the relation between adjacent turns is always
mediated by an evaluative assessment as a precondition for the production of
a succeeding turn. How constituent moves are assessed by participants is
crucial to the production of conversational exchanges. The object of IS analy-
sis is thus not a ready-made sequence of turns at speaking. From an interac-
tional sociolinguistic perspective, exchanges are best represented as sequences
such as the following: ‘move … evaluative assessment … countermove …
evaluative assessment … further move’, and so on. That is to say, any ex-
change depends on participants’ interpretive acts at any one of the many
transition points in the exchange. Apart from the usual semantic processes,
moreover, ideology, which in turn closely reXects societal power relationships,
also enters into the assessments. In all these matters, IS analysis clearly diVers
from conversation analysis.
Eerdmans’ two examples (this volume: 90V) illustrate these matters. In the
second case, where background assumptions are shared, each of the two
principals can rely on shared knowledge of indexical conventions to indirectly
convey conXicting accounts of what transpired. In the Wrst example, the job
center interview, the two participants diVer in background knowledge and are
unsuccessful in establishing a productive exchange, with the result that the
interviewer is unable to make a fair assessment of the applicant’s abilities.
Because of the prevailing ideology of minorization, and what are in fact
diVerences in background knowledge, the applicant’s abilities are disvalued
and his career prospects damaged. This one exchange clearly reveals how
macro-societal stereotypes can aVect our assessments of what transpires in
such encounters.
Ballim’s brief comments are helpful because they point out speciWc areas
of application where the interactional sociolinguistic approach may be useful.
The comments highlight areas of interest where IS and Natural Language
Processing converge. Among other things, the two share an interest in situ-
ated interpretation as it aVects participants’ perception of context. But as an
anthropologist, I am inclined to place more emphasis on shared traditions and
how they aVect speaking practices. For instance, building on Wndings from
work on intercultural communication, I have tried to isolate shared interpre-
tive principles or maxims demonstrably grounded in the typiWcations that
arise from participation in speciWc networks of relationships. Such maxims
Response essay 119
must of course ultimately be shown to be veriWed in the conduct of interper-

sonal relations.
With respect to indirect reasoning, IS analysis could serve as an initial
empirical heuristic that could test a priori assumptions about the kinds of
reasoning processes children would assume to be shared and what they would
assume to be explicitly lexicalized. This is a distinction that has been useful in
my own work on classroom talk, where I often had to deal with diYculties in
teacher-student communication.
As to the other topics Ballim brings up, inference, as I use the term, clearly
involves indirect reasoning and the distinction between lexicalized and im-
plicitly conveyed matters is basic. I often use Harold GarWnkel’s term “taken-
for-granted knowledge” to refer to such matters. In the studies of classroom
teaching processes, brieXy referred to at the beginning of the discussion,
English teachers, as well as their ten-year-old students, all talked about gram-
matical issues that came up in reading comprehension. But whereas the teach-
ers relied on explicitly lexicalized grammatical terminology to make their
point, students, when talking, dealt with grammar almost entirely in indirect
terms. They had diYculty with the teachers’ grammatical terminology
(Gumperz and Field 1995; Gumperz, Cook-Gumperz and Szymanski 1999).
I am grateful to Ballim for the way he characterizes my position on
community. My main point is, as he puts it, that “it is not suYcient to consider
discourse in terms of a uniform, disembodied community.” What we must
account for is individual action as reXected in practice, but I Wnd the term
“rhetorical tradition” more useful in accounting for the fact that whatever we
know or believe is acquired through participation in some collectivity. I have
used the term “event” in two senses. First, historically, in describing the
ethnographers of communication’s shift away from community-wide ethnog-
raphy. Secondly, as a heuristic in IS Weldwork on communicative practice.
With respect to context, psychologists, cognitive scientists, and many
linguists who pay attention to context tend to deWne it almost entirely in extra-
communicative terms. I argue that, while these factors are, of course, sig-
niWcant, contextual information is imported into the interpretive process
primarily via indexical contextualization cues, in the form of presuppositions
of what the activity is and what is communicatively intended. More compara-
tive work on inferential processes needs to be done to test out what are at the
moment very tentative generalizations. The little comparative work I have
done so far indicates that the indexical and inferential processes readily circu-
late across language boundaries, which is another argument for distinguishing

between language and communicative practice. But how they circulate is still
far from clear.
Finally, on the matter of what Prevignano in our discussion (this volume:
21V), and later Ballim in his commentary (this volume: 83) have called “cor-
pus pragmatics”: one of the obstacles to valid corpus-based comparative
research is the lack of agreement on what is analytically signiWcant data.
Particularly in pragmatics, we Wnd wide disagreements on what constitutes
valid data and how it needs to be transcribed and formatted for further
analysis. In my recent work, I have used Michael Silverstein’s notion of
“interactional text” (1993), which my co-author and I deWne as a transcription
that sets down on paper “all those perceptual cues that past research and
ongoing analyses show participants rely on in their online processing of
conversational management signs” (Gumperz and Berenz 1993:92).1 The
current discussions of “entextualization” as the process by which mental
processes are translated into words should also be relevant here (Silverstein
and Urban 1996).
Thibault, in his contribution, critically examines the IS approach to com-
municative practices, comparing it to an alternative which, like the IS
approach, assumes multimodality of signaling processes, but where com-
municative practice is analyzed with reference to a “community’s resource
systems.” Seen from my perspective, locating shared linguistic resources in “a
community” raises serious problems. To speak of a community’s resource
systems implies a deWnition of what is meant by the term. Currently, uses vary
from the traditional treatments of community as an abstract, disembodied,
uniform whole, to notions like “communities of practice” (Lave and Wenger
1991), that is, sets of individuals regularly cooperating in the accomplishment
of some set of tasks. For example, cognitive psychologist Herbert Clark (1996)
deWnes community as any set of individuals who share some form of expertise:
residential groups, professional groups, voluntary organizations and the like.
It is unclear, apart from lexical labels, how these deWnitions are reXected in
language. Traditional deWnitions of community, on the other hand, are, as we
all know, becoming more and more controversial in today’s increasingly
diverse communicative environments. In sum, I am far from clear about how
the concept of community can be integrated into an analysis of communica-
tive practices. My position is that more comparative research on the distribu-
Response essay 121
tion of shared linguistic and social knowledge is needed before we can resolve
this issue (see also Gumperz 1997).
Towards the beginning of his discussion of context and context-indepen-
dent meanings, Thibault refers to some of my statements that he interprets as
contrasting “pure indexicals with context-independent or propositional
meaning.” Perhaps my use of the term “propositional” and “context-free” is
confusing, but that is not what I intended to say. I agree with Thibault that all
naturally-occurring semiotic forms are indexical. To the extent that we can
talk about propositional content and context-independent meanings, we refer
to forms or interpretive assessments that have been extracted from some
context by analysts for comparative or other analytical purposes. I should
have made that clearer.
But other sections of Thibault’s comments reXect fundamental diVer-
ences in basic assumptions with respect to those on which IS rests. For
example, he argues that communicative practices have to be studied in rela-
tion to a community’s resource systems. These are described in terms of arrays
of semiotic modalities, including grammar, prosody, non-verbal signs etc., in
fact pretty much the same types of signs that IS treats as entering into the
inferential process. But whereas for IS the shared inferencing of human actors
is the locus of interpretive assessment and ultimately understanding (or
“meaning-making”, in Thibault’s terms), Thibault argues that meaning
construal is achieved by a conjunction of mutually contextualizing modalities.
In other words, meaning, on this view, resides in language, whereas IS follows
the linguistic anthropological tradition in arguing that it is humans who
construe meaning (Lucy 1993). Both approaches have overcome the limita-
tions of traditional approaches that see only denotation as properly linguistic.
But beyond that, I have problems with Thibault’s position. How does he
deWne a community in today’s conditions of ever-increasing diversity? How
does he account for the human factor in interpretation, particularly its ideo-
logical character and limited access to knowledge? Through IS analysis it is
possible, for example, to gain insights into how communicative barriers arise;
what it is that brings about their disappearance; how diversity can be studied
via communicative practices in ways that do not rely on a priori assumptions
about group or community membership; how interpretation reXects past
communicative experience. I believe that I have reason to claim that IS has, at
the least, made signiWcant steps towards the solutions of the problems I set
myself and I am not sure how Thibault would deal with such issues. But then
the two approaches reXect diVerent interests and cannot be judged by their
inability to deal with matters they were not designed to deal with.
Elsewhere in this volume, I have already partially answered Prevignano’s
questions about what makes interaction possible and about the validity of
inference-based interpretive assessments. Let me brieXy recap: I see interac-
tion as a turn-by-turn process, where the production of each turn, including
the opening move, is preceded by an assessment that is simultaneously prag-
matic and meta-pragmatic. Interaction always presupposes conversational
involvement, so that the maintenance of involvement depends, among other
things, on the extent to which knowledge of linguistic, that is, symbolic and
indexical signs, and underlying background knowledge are shared.
Both interactants and analysts are thus always faced with ambiguities. But
the resulting interpretive diYculties seem less problematic if, instead of look-
ing at interpretation as a one-shot assessment, we think of it as an ongoing
interactive process where inferences are constantly revised and attuned as
circumstances change.
As to the question of minims and maxims, Prevignano (this volume: 63),
if I interpret him correctly, is comparing two scholarly positions, the Gricean
tradition, which postulates universal conversational principles or maxims at
the level of content, and a second one that he associates with Gumperz. The
latter argues that interpretive assessment also depends on context-speciWc
aspects of surface discursive form that do not ordinarily enter into other types
of interpretive analysis. I Wnd Prevignano’s discussion interesting because he
seems to be attempting to deal with issues of appropriate speech that many
pragmaticists and sociolinguists tend to account for by means of the much
criticized normative rules of appropriateness, of the form: “In situation A, do
or use Y”, where the term “situation” refers to extra-communicatively avail-
able sets of circumstances. The two traditions he alludes to both take a basi-
cally semiotic position on what makes for rhetorically eVective talk,
attempting to capture regularities of discursive practice without reifying them
in terms of a priori rules. The two diVer in that whereas the Wrst tradition
concentrates on content, the second argues that interpretation rests on par-
ticipants’ ability to relate form to content both to decode propositional mean-
ing and to derive contextually-grounded hypotheses about presuppositions
and possible entailments that aVect understanding. My insights into the com-
municative import of linguistic form derive from work in dialectology and
intercultural communication, where what from a structural perspective
Response essay 123
seemed like insigniWcant phonetic and prosodic processes, can have great
communicative import. I believe that to account for the facts of discourse-
level interpretive assessments, we need to treat discursive or communicative
practice as a separate level of analysis.
Prevignano brings up the issue of metacommunication in discussing the
validity of analysts’ models or descriptions. Let me point out, however, that
one of the main reasons linguistic anthropologists, building on Voloshinov’s
discussion in Marxism and the Philosophy of Language (1973), have begun to
concentrate on meta-communicative phenomena is to avoid their prede-
cessors’ reliance on analytically unjustiWed, dichotomous distinctions like
‘language and thought’, ‘language and culture’, and the like. Linguistic anthro-
pologists have turned to metacommunication as a strategy for avoiding these
dichotomies. In so doing, they open up new possibilities as well as empirical
methods for dealing with these issues. The reason for these new approaches is
the realization that the way people talk about talk provides at least an indirect
indication of what people might think about what they hear. In other words,
metacommunicative phenomena provide us with benchmarks for evaluating
analysts’ interpretations. The distinctions between explicitly metacommu-
nicative signs and indirect indexicals such as contextualization cues is, ana-
lytically, of great importance. The fact that the former are readily subject to
conscious manipulation, whereas the latter are automatically produced below
the level of consciousness, opens up interesting possibilities for interpretive
analysis, as several recent articles suggest (Gumperz 1999; Cook-Gumperz
and Gumperz 1996).
Finally, as to my current interests, at the moment I am working on a book
to be entitled Language in Social Theory, which reexamines Wndings from my
Weldwork over the years in the light of recent theoretical developments, ex-
panding, among other things, on the ideas expressed here. In addition, I am
about to Wnish a case study, Communicative Competence as Communicative
Practice, which reviews the notion of communicative competence in the light
of what we have learned about discourse since the seventies. I am also cooper-
ating with Marco Jacquemet on an edited volume of ethnographically-based
articles that will pick up where Gumperz and Hymes’ Ethnography of Commu-
nication (1972) left oV.
Note
1. See also the article by Du Bois et al. in the same volume (1993), and the more recent, more
comprehensive, “Gesprächsanalytisches Transkriptionssystem (GAT)” (Selting et al.,
1998).
References
Ballim, Afzal. “A commentary on a discussion with John J. Gumperz”. This volume.

Blom, Jan-Petter and Gumperz, John J. 1972. “Social meaning in linguistic structure:
Code-switching in Norway”. In Directions in Sociolinguistics: The Ethnography of
Communication, J. J. Gumperz and D. Hymes (eds.), 407–434. New York: Holt,
Reinhart and Winston.
BloomWeld, Leonard. 1933. Language. New York: Holt, Rinehart and Winston.
Clark, Herbert H. 1996. “Communities, commonalities and communication”. In Rethink-
ing Linguistic Relativity [Studies in the Social and Cultural Foundations of Language
17], J.J. Gumperz and S.C. Levinson (eds.), 324–355. Cambridge: Cambridge Univer-
sity Press.
Cook-Gumperz, Jenny and Gumperz, John J. 1994. “The politics of a conversation: Con-
versational inference in discussion”. In What’s Going on Here? Complementary Stud-
ies of Professional Talk, A.D. Grimshaw et al. (eds.), 373–397. Norwood, NJ: Ablex.
Cook-Gumperz, Jenny and Gumperz, John J. 1996. “Treacherous words: Gender and
power in academic assessment”. Folia Linguistica XXX (3–4):167–188.
di Luzio, Aldo. “Presenting John J. Gumperz”. This volume.
Du Bois, John W. et al. 1993. “Outline of discourse transcription”. In Talking Data:
Transcription and Coding in Discourse Research, J.A. Edwards and M.D. Lampert
(eds.), 45–89. Hillsdale, NJ: Lawrence Erlbaum.
Edwards, Jane A. and Lampert, Martin D. (eds.). 1993. Talking Data: Transcription and
Coding in Discourse Research. Hillsdale, NJ: Lawrence Erlbaum.
Eerdmans, Susan L. “A review of John J. Gumperz’s current contributions to Interactional
Sociolinguistics”. This volume.
Gomez-Imbert, Elsa. 1996. “When animals become “rounded” and “feminine”: Concep-
tual categories and linguistic classiWcation in a multilingual setting”. In Rethinking
Linguistic Relativity [Studies in the Social and Cultural Foundations of Language 17],
J.J. Gumperz and S.C. Levinson (eds.), 438–469. Cambridge: Cambridge University
Press.
Gumperz, John J. 1971. Language in Social Groups. Stanford: Stanford University Press.
Gumperz, John J. 1992a. “Contextualization and understanding”. In Rethinking Context:
Response essay 125
Gumperz, John J. 1992b. “Contextualization revisited”. In The Contextualization of Lan-

guage, P. Auer and A. di Luzio (eds.), 39–53. Amsterdam and Philadelphia: Ben-
jamins.
In Rethinking Linguistic Relativity [Studies in the Social and Cultural Foundations of
Language 17], J.J. Gumperz and S.C. Levinson (eds.), 374–406. Cambridge: Cam-
Gumperz, John J. 1997. “On the interactional bases of speech community membership”.
In Towards a Social Science of Language: Papers in Honor of William Labov, vol. II:
Social Interaction and Discourse Studies, G.R. Guy et al. (eds.), 183–203. Amsterdam
and Philadelphia: Benjamins.
Gumperz, John J. 1999. “Culture in the cultural defense”. Proceedings of the Sixth Annual
Symposium about Language and Society (1998), SALSA VI:115–132. Austin: Depart-
ment of Linguistics, University of Texas.
Gumperz, John J., Cook-Gumperz, Jenny and Szymanski, Margaret. 1999. Collaborative
Practices in Bilingual Cooperative Learning Classrooms. CREDE Research Report 7.
Santa Cruz, CA and Washington DC: Center for Research on Education, Diversity
and Excellence.
Gumperz, John J. and Field, Margaret. 1995. “Children’s discourse and inferential prac-
tices in cooperative learning”. Discourse Processes 19:133–147.
Gumperz, John J. and Hymes, Dell (eds.). 1972. Directions in Sociolinguistics: The Ethnog-
raphy of Communication. New York: Holt, Reinhart and Winston.
Gumperz, John J. and Levinson, Stephen C. (eds.). 1996. Rethinking Linguistic Relativity
[Studies in the Social and Cultural Foundations of Language 17]. Cambridge: Cam-
Hanks, William F. 1996. Language and Communicative Practices. Boulder, CO: Westview
Press.
Lave, Jean and Wenger, Etienne. 1991. Situated Learning: Legitimate Peripheral Participa-
tion [Learning in Doing: Social, Cognitive, and Computational Perspectives]. Cam-
Levinson, Stephen C. “Contextualizing ‘contextualization cues’”. This volume.
Lucy, John A. 1993. “ReXexive language and the human disciplines”. In ReXexive Lan-
guage: Reported Speech and Metapragmatics, J.A. Lucy (ed.), 9–32. Cambridge: Cam-
Prevignano, Carlo L. “On Gumperz and the minims of interaction”. This volume.
volume.
Selting, Margret et al. 1998. “Gesprächsanalytisches Transkriptionssystem (GAT)”. Lin-
guistische Berichte 173:91–122.
enough?” In The Contextualization of Language, P. Auer and A. di Luzio (eds.), 55–

76. Amsterdam and Philadelphia: Benjamins.
Silverstein, Michael. 1993. “Metapragmatic discourse and metapragmatic function”. In
ReXexive Language: Reported Speech and Metapragmatics, J.A. Lucy (ed.), 33–58.
Silverstein, Michael and Urban, Greg. 1996. Natural Histories of Discourse. Chicago and
London: University of Chicago Press.
Thibault, Paul J. “Contextualization and social meaning-making practices”. This volume.
Body dynamics, social meaning-making, and scale heterogeneity 127
Chapter 9
Body dynamics, social meaning-making,

and scale heterogeneity
Re-considering contextualization cues
and language as mixed-mode semiosis1
Paul J. Thibault
1. Language as mixed-mode semiosis: Topological and typological

dimensions
Gumperz makes the distinction between “contextualization cues”, which are a

class of pure indexicals, and “symbolic signs”, including symbolic lexical and
grammatical signs. According to Gumperz, contextualization cues “have no
propositional content” (Prevignano and di Luzio, this volume: 9). Unlike
symbolic lexical and grammatical signs, contextualization cues qua pure
indexicals “signal only relationally and cannot be assigned context-free lexical
meanings” (Prevignano and di Luzio, this volume: 9). Levinson makes a
similar distinction between “conscious ‘verbal’ coding and less conscious, less
fully-coded prosodic and paralinguistic channels” (Levinson, this volume:
36). The former are explicit and fully foregrounded; the latter tend to be
vague, less explicit, non-propositional and backgrounded (see Levinson, this
volume: 36). “Contextualization cues” provide the interpretive background
for foregrounded explicit propositional meaning.
In this chapter, I shall explore some further implications of distinctions
proposed by Gumperz and Levinson in order to explore how we can develop
a more integrated account of the relations between implicit and explicit
modes of semiosis. Both implicit and explicit modes are intrinsic to the
internal organization of language. Both are necessary for explaining how
language functions in contexts which implicate semiotic processes on a di-
versity of space-time scales that go beyond the here-now scale of the given
occasion of discourse.
It is interesting to note, in the light of the observations made in the

preceding paragraph, how the distinction between formally segmentable
propositional meanings and the holistic and non-propositional properties of
contextualization clues reXects the entrenched separation in much of the
language sciences of semantic-propositional meanings as explicitly coded in
lexicogrammar, and other meanings of the kind grouped under the term
“contextualization cue” as being extrinsic to language form, i.e. as impinging
upon or otherwise aVecting the latter from outside (Thibault and Van
Leeuwen 1996). The problem with this account is that it preserves a view of
language in which segmentable propositional forms are de-contextual and
unvarying from one context to another, standing independently from the
subjective perspectives of language users and the richly textured and situated
nature of language-in-use. In this view, the core of linguistic organization
stands apart from the eVects of time-bound contextualization as abstract and
unchanging formal features.
The solution to this problem lies in a framework which recognizes that
semiotic systems, including language, are dynamic open systems which are
cross-coupled to both the ecosocial and bodily environments of their users.
Language, along with its ecosocial and bodily environments, are continually
interacting with each other in the real-time of discourse events. In the process,
language entrains bodily and ecosocial dynamics to its own dynamics. The
distinctions between, for example, explicit propositional and implicit non-
propositional meanings suggest that language can be divided into indexical
and symbolic components and functions. However, in discourse, such dis-
tinctions only emerge and co-evolve as the overall trajectory of the discursive
event unfolds in time.
In such a view, context is not something which is external to and indepen-
dent of the intrinsic properties and dynamics of semiotic systems. But this is
exactly the view which the separation between explicit, formally segmentable
proposition forms and contextualization cues has not entirely succeeded in
shaking oV. This is so, I argue, because the former notion preserves the idea
that some properties of language exist independently of context, whereas
others are contingent contextualization cues which act on the former without
being essential to the explanation of the internal organization of language
form. However, it is the fact that both formal and functional properties
emerge as a result of the embeddedness, so to speak, of language in context
which should be taken as a central fact about language. This requires an
alternative view, in which the distinction between context-independent and

context-dependent meanings and forms gives way to one in which both
proposition-bearing forms and contextualization cues, to stay with the
Gumperzian parlance for the moment, are seen as depending on the overall
ecosocial context in which the various components interact with and are
integrated with each other.
In order to accomplish this task, it is necessary to investigate language as a
mixed semiotic mode, displaying both typological-categorial and topological-
continuous modes of organization. In this regard, it is useful to reXect on the
following quotation from Lemke (1999), who distinguishes two main classes
of semiosis along these lines, as follows:
…(a) those cases in which the features of representamen that are criterial for
some SI [system of interpretance, PJT] to interpret it as a sign of some X may
vary continuously, so that quantitative diVerences of degree in a feature of R
normally lead to diVerences of degree or kind in the interpretant, vs. (b) those in
which all representamina are classiWed by the SI into a discrete spectrum of
types, and each R-type is interpreted as a distinct X.
(Lemke 1999: 9)
The Wrst type is topological-continuous variation; it is quantitative diVerences

of degree that makes a diVerence in this mode. The second type is founded on
discrete typological-categorial distinctions. The Wrst type is a generalization of
Bateson’s notion of “analogic communication”; the second generalizes his
notion of “digital communication” (Bateson 1973 [1972]:342–344). The
lexicogrammar and semantics of natural language are, for the most part, based
on typological-categorial diVerence. Lemke further points out that while it is
possible to “map continuous variation in X onto continuous variation in R
(and vice versa),” (topological) “discrete variants of X onto discrete variants of
R (and vice versa),” (typological) there may also be “mixed modes of semiosis
in which the continuous is mapped onto the discrete and vice versa.” These
mixed modes are fundamental to the understanding of the dynamics of multi-
scale dynamical systems such as language.
Language is often made to appear to be less of a mixed semiotic mode than
I believe it actually is. This is so because of the focus on explicit propositional
meaning qua symbolic representational mode, along with the formally seg-
mentable lexicogrammatical forms which realize this, seen as the deWning
characteristic of language. This focus emphasizes the predominantly typologi-
cal-categorial nature of lexicogrammatical forms and the semantic categories
which these forms realize. This means that both the semantic categories and
the lexicogrammatical forms which realize them are discrete. That is, the
categories and the forms which realize them are deWned by contrasts with
other discrete forms and categories in relation to the paradigmatic systems
and the structures whereby language is theorized as a system of contrasting
terms in which their respective values are deWned by the place of each term in
the overall system. For example, consider the clause complex, When tides are
very large, Wsh in the estuaries for golden snapper, trevally, salmon, cod and the
elusive barramundi. The second clause in this complex is an instance of the
grammatical category ‘imperative mood’. Imperative mood is usually seen as a
discrete type-category which corresponds with other mood categories such as
declarative, interrogative, exclamative moods in the mood system of English
grammar. The noun tides also illustrates the same point. This noun is plural
rather than singular, as indicated by the morphemic suYx -s. The noun is
either plural or singular; the grammar does not recognize continuously vary-
ing degrees of plurality or singularity between these two categorial dis-
tinctions. This is a discrete typological-categorial distinction which the
lexicogrammar of English realizes through, for example, the presence or ab-
sence of morphemes such as the one mentioned here. English grammar has no
resources for specifying topological-continuous variation between the seman-
tic distinctions [singular] and [plural].
Furthermore, the nouns tides, estuaries, snapper, trevally, salmon, cod
and barramundi specify discrete type-categories of the schematic category
Thing, which is prototypically realized by nouns in English. Thus, a given
Wsh which the fortunate angler catches in the estuaries around Darwin Har-
bour is likely to be categorized as an instance of the semantic type-category
represented by the nouns snapper, trevally, salmon, cod and barramundi.
These words refer to discrete categories of Wsh, just as the lexicogrammatical
forms qua exponents of the word class ‘noun’ are themselves discrete: the
noun snapper names the semantic category [snapper] rather than the cat-
egories [salmon] or [barramundi]. Nor does this noun name continuously
varying degrees of ‘snapperness’ or ‘salmonness’ or ‘barramundiness’ which
lie somewhere between these discrete distinctions. The typological mode of
semiosis is certainly a prominent characteristic of language. Nevertheless,
language is not exclusively organized according to typological-categorial
principles. As the observations made by Gumperz and Levinson on
contextualization cues show, language also exhibits topological-continuous
modes of semiosis which are integrated with the typological-categorial prin-

ciples outlined above. In other words, language has properties of mixed-
mode semiosis and this requires that both its internal organization and
contextual functioning be explained accordingly.
The observations made in the preceding two paragraphs may also be
extended to the phonology of the language. In the Indo-European tradition
of linguistics, the longstanding focus on the phoneme has privileged seg-
mental forms of phonological organization, along with the function of pho-
nemes and syllables, in symbolizing categorial distinctions such as
morpheme and word units and boundaries on the lexicogrammatical level.
Yet, the contextualization cues that Gumperz focusses on have to do with
how “the phonological Xow manages the Xow of discourse” (Halliday
2000a:108). The distinctions which phonemes and syllables make between
this and that morpheme or word are typological-categorial ones. Further-
more, they reXect that aspect of the phonological system which arbitrarily
realizes lexicogrammatical distinctions. Firth’s critique of such a mono-
systemic approach to phonology had long ago envisaged a polysystemic
science of the sounds of language as involving “a plurality of systems of
interrelated phonematic and prosodic categories” (1957 [1948]:137). The
role of the phonological Xow in the Xow of discourse has to do with
nonarbitrary relations between phonology, on the one hand, and lexico-
grammar and discourse, on the other. In English, for example, rhythm and
intonation are central here. However, both rhythm and intonation are in-
trinsic to the internal organization of English phonology, such that phonol-
ogy, on analogy with lexicogrammar, is seen as comprising a number of
domains of systemic organization deWned as a rank scale of hierarchically
organized units. In English, these phonological units are a tone unit, a
rhythm unit (the foot), and an articulatory unit (the syllable) (Halliday 1992,
2000a). In English, nonsegmental modes of interpersonal meaning (e.g.
mood and modality) and textual meaning (e.g. the organization of discourse
as Given and New information) in lexicogrammar are often integrated with
nonsegmental features of rhythm and intonation in discourse. Therefore,
such “contextualization cues” in English phonology are not external to lan-
guage design; rather, they are nonsegmental forms of phonological organiza-
tion which are constitutively inseparable from the segmental or particu-
late forms of organization which are also found in both phonology
and lexicogrammar.
The tripartite perspective on language as “particle”, “wave”, and “uncoded

Weld”, Wrst proposed by Pike (1967) and subsequently taken up and adapted
by Halliday (1979), shifts the emphasis away from the distinction between
fully-coded propositional content and contextualization cues. Instead of this
distinction, Halliday’s adaptation of Pike’s tripartite perspective proposes that
a number of diverse functional regions, or metafunctions, are simultaneously
conWgured in lexicogrammatical form. Halliday’s (e.g. 1979) metafunctional
account claims that the intrinsic functional organization of lexicogrammatical
form is organized in terms of four general semantic regions, each having their
characteristic modes of formal organization in the lexicogrammar of natural
language. According to Halliday, lexicogrammar simultaneously: (i) catego-
rially construes experience, including naming and referring; (ii) enacts inter-
personal relationships; (iii) provides the means whereby language coheres
textually as discourse which is operational in context; and (iv) connects
linguistic units to each other in relations of logical (causal, temporal, etc.)
dependency. These four distinct functional regions are referred to in syste-
mic-functional linguistic theory as the “experiential”, “interpersonal”, “tex-
tual”, and “logical” metafunctions, respectively.
Each of the four metafunctions favors a diVerent mode of formal (lexico-
grammatical) realization. Thus, experiential meanings are realized by particu-
late or segmental forms of realization, textual meanings by wave-like or
periodic ones based on the ways in which waves of discursive activity Xow into
and merge with each other, and interpersonal meanings favor prosodic or
scopal forms of realization such that an interpersonal Weld is the domain
surrounding a given interpersonal selection — e.g. mood or modality in the
clause. In this perspective, the given interpersonal selection (e.g. mood or
modality in the clause) syntagmatically extends over and in some way
modiWes or deforms (inXuences) some other feature within the given
syntagmatic Weld which it holds in its scope. In so doing, the interpersonal
feature inXuences and shapes the given syntagmatic Weld for a particular
interactive purpose. Pike’s original proposal thus allows us to see that lan-
guage is internally organized according to a number of diVerent, yet comple-
mentary, perspectives, such that no single perspective such as the particulate
or segmental one which has dominated linguistics can satisfactorily explain all
aspects of the formal organization of language. Experiential, textual, and
interpersonal meanings are nonarbitrarily-related to their diVerent modes of
lexicogrammatical realization such that both typological-categorial and to-

pological-continuous modes of semiosis are seen to operate within lexico-
grammatical form, simultaneously. Parallel principles of organization also
operate on the expression stratum of phonology, as Firth (1957 [1948])
showed, such that the nonsegmental dimensions of linguistic organization are
fully integrated with the segmental ones. In such a perspective, the distinction
between segmental propositional forms and nonsegmental contextualization
cues can be assimilated to one in which the nonarbitrary or natural relation
between phonological processes, the human body, lexicogrammatical and
discourse processes is revealed as fundamental to the nature and functioning
of language.
This requires a perspective which sees language as organized on diverse
space-time scales. In the next section, I shall explore this question.
2. The semiotic Principle of Alternation and mixed-mode semiosis
Lemke has proposed the Principle of Alternation as a way of theorizing, in

terms of the three-level scalar hierarchy, how semiotic functions are mapped
onto dynamical scale levels, and the re-organization of continuous variation
into discrete variants. Lemke deWnes this Principle as follows:
Each new, emergent intermediate level N in a complex, hierarchical, self-organiz-
ing system functions semiotically to re-organize the continuous quantitative
(topological) variety of units and interactions at level (N–1) as discrete, categorial
(typological) meaning for level (N+1), and/or to re-organize the discrete, cate-
gorial (typological) variety of level (N–1) as continuously variable (topological)
meaning for level (N+1).
(Lemke 1999: 9; italics in original)
Gumperz’s discussion (Prevignano and di Luzio, this volume: 9–10) of the

interaction between symbolic signs with propositional content and indexical
signs functioning as contextualization cues suggests the mixed semiotic
modes mentioned above. This brings into focus two complementary perspec-
tives on semiosis. The Wrst, topological mode is grounded in the materiality of
our embodied being-in-the world interactively. It is closer to our primary
bodily being-in-the-world before language. Its scale is, in the Wrst instance, the
organismic one of the body-brain in its immediate physical-material environ-
ment. The second mode represents the system’s expansion into typological-
categorial phase space whereby semiosis cascades and ramiWes across space-
time scales not tied to the here-now scale of the organism’s material
interactivity with its immediate environment (Lemke 2000:194; Thibault
2000). Whereas the topological mode is more tightly cross-coupled to the
material processes of the body, the typological mode predisposes the linking
of a particular meaning-making event to space-time scales beyond the here-
now scale of the organism’s material interactivity with its immediate environ-
ment. This does not mean that the typological mode transcends the
topological mode. The point is that language-in-use, rather than language as
abstract system, operates on the basis of “mixed-mode” semiosis such that it
combines both topological and typological principles as central to its contex-
tual functioning.
The mapping of semiotic functions onto diverse organizational scales has
dynamical implications for the emergence of meaning, viz. the emergence of a
new level in the scalar hierarchy of semiotic functions can only occur, both
phylogenetically and ontogenetically, if and only if a new level in the hierarchy
of semiotic interpretance emerges. Each higher level in the hierarchy has
multiple realizability at lower levels. For example, it is the many-to-one map-
ping of topological-continuous variety at the N–1 level of the continuous
topological Xux of the phenomena of experience onto the N level of
lexicogrammar that allows for the Wltering of N–1 non-criterial Xuctuations at
the N level. Furthermore, discrete lexicogrammatical forms at this level can be
re-construed at the N+1 level as having an increased variety of meanings,
depending on the ways in which constraints on the N level of lexicogrammar
are taken up and responded to on the basis of their connections to a culture’s
intertextual networks on the more global, longer term N+1 level of the SI (see
Lemke 1999; Thibault, forthcoming).
Following Lemke’s Principle of Alternation, we see that the semiotic trans-
formation of continuous topological variation in the Xux of perceptual and
other phenomena of experience to discrete lexicogrammatical forms and of
these to continuous semantic meanings, in the progression from level N–1 to
level N+1, constitutes what Lemke deWnes as “a semiotic transformation of the
information content of lower levels as signs for higher levels, allowing many-
to-one classiWcations and one-to-many context-dependent reinterpretations”
(1999: 9). It is the dynamics of the semiotic transformations mapped onto the
scalar hierarchy that allows for the emergence of the symbolic possibilities of
language, and hence its capacity to form contextual links with many diverse
space-time scales in a given act of meaning-making.
3. Material and semiotic dimensions of meaning-making
Meaning-making is both a physical-material process and a semiotic-discur-

sive one. Gumperz emphasizes the social dimension of meaning-making with-
out, however, giving equal importance to the material dimension. The
material and social dimensions are equally necessary to the development of an
adequate theory of how meanings are made in a given ecosocial system. Social
practices are also material processes. Gumperz emphasizes the social practices
through which meaningful relations are jointly constructed in and through
the social activities whereby agents interact with each other. The material
dimension of meaning-making brings it into contact with other physical-
material processes and Xows, including biological ones, in the ecosocial sys-
tem. In meaning-making, semiotic and material processes are cross-coupled
as meaning-making activity. It is such networks of activities which have
spatial and temporal extension beyond the local here-now that support and
sustain particular cultural practices. There is no question of a localized, homo-
geneous community deWned on the basis of one’s membership of a given
language, social dialect, and so on. Networks of human activity are not neces-
sarily localized in this sense. The point is that symbolic meaning-making
resources in particular enable human agents, material objects, and so on to
interact with each other across very many diVerent space-time scales qua
social-cultural network. They do so on the basis of their participation in the
meaning-making practices in and through which a given network is deWned. I
want to make clear that the deWnition of network oVered here diVers in several
ways from Gumperz’s use of this term.
According to Gumperz, “community membership is of course directly
linked to participation in such networks of relationships, but in our post-
industrial worlds, it is less and less possible to take this for granted”
(Prevignano and di Luzio, this volume: 18). Gumperz directly equates “com-
munity membership” with participation in “institutionalized networks of
relationship”, such that networks are seen as localized to speciWc communities
in space and time, presumably according to (unspeciWed) assumptions made
about the pre-industrial world. In the deWnition I am operating with here, a
network is the organization of material and social processes across potentially

diverse space-time scales. In this way, the problem of the localization of a
community and its reduction to linguistic membership gives way to that of
which cross-couplings of material and social processes sustain the social-
cultural practices of the network. Such a network cannot exist independently
of such cross-couplings. Ecosocial networks have a “1-dimensional reticular
topology spread through a three-dimensional ecosystem” (Lemke 2000:200).
Networks do not have deWnite edges or boundaries. For this reason, they
cannot, by deWnition, be reduced to speciWc point locations.
In the view outlined above, a language system’s internal organization
consists in the speciWc terms and the diVerential relations among these. How-
ever, a system also has higher-scalar environmental constraints or boundary
conditions that interact with the system at any given moment in time such
that the former can aVect and shape the latter. In such a view, context-
invocative, aVectual, rhetorical, subjective, and so on, properties of language-
in-context are neither epiphenomenal nor secondary. Moreover, the notion of
a “language system” does not refer to some reiWed entity ‘out there’ which
exists independently of the social meaning-making practices of a given social
group and which has an independent causal status. Instead, the notion of the
language system refers to the semiotically salient diVerences which the mem-
bers of a given social group recognize as potentially meaningful. These
diVerences are immanent in the meaning-making practices of the members of
some social group as well as being immanent in the symbolic neural space of
biological individuals.
The causal status of such a system of diVerences is not explainable in
externalist terms as if the language system, so deWned, causally acted upon and
hence determined the meaning-making behavior — the linguistic choices, and
so on — of social agents. It is not a question of a mechanistic cause-and-eVect
view of grammar as in, say, Chomsky’s (1965) transformational-generative
model. In such a view, grammar qua competence and individual performance
stand in a cause-and-eVect relation within the same scalar level of the indi-
vidual organism. Rather, it is a system of constraints or boundary conditions
whereby higher scalar levels informationally constrain lower levels. The
semiotic activities of the social group is not a mere aggregate of the behaviors
of the individual organisms. This applies to all forms of semiotic activity such
as, for example, language, gesture, depiction, and so on. These cannot be
reduced to or uniquely explained in terms of the neurophysiology of indi-
vidual organisms or the locally perceived goals and intentions of social actors.
Rather, the behaviors of the individual organisms involved are entrained to
higher-scalar ecosocial patterns.
Furthermore, the organization of the higher-scalar level cannot be de-
scribed in terms of the organization of entities and processes existing on
lower scalar levels. For example, Cowley (1998) has shown how the interper-
sonal attunement of individuals to each other in conversational events can
depend on physical properties of utterances such as the rate at which partici-
pants vibrate their vocal chords as they tune in to and synchronize their
voice with variables such as Xuctuations in fundamental pitch of the other’s
voice (1998:549). Such sensori-motor events take place on a temporal scale
that is much faster than the rising and falling tones that are recognized as
meaningful contrasts in intonation. Intonational contrasts are perceivable
and are salient on a slower temporal scale than that of the centiseconds
characteristic of the micro-temporal scale of vocal chord vibration. Sono-
graph records show, as Cowley (1998:550–551) documents, that participants
in conversation tune into and act on each other at this physical acoustic level
of, for example, pitch Xuctuation. The diVerent temporal scales of the enti-
ties and processes on the lower scalar sensori-motor level as compared to
that of the higher-scalar level where contrasts in intonation are salient shows
how, in speaking, participants both act on and respond to each other on the
basis of voice dynamics existing on potentially many diVerent temporal
scales which implicate physical-material (bodily) processes not normally
considered to be intrinsic to linguistic form, but which are acted on by
participants as meaningful components of the act of conversing with others.
The process whereby participants converge on a particular fundamental
pitch frequency is an example of the embodied ways in which interactants
tune into each other in ways recognized as interpersonally salient.
In other words, the very fast, lower scalar vocal dynamics of individual
participants are entrained to higher-scalar rhythmic and harmonic patterns at
the level of the conversational event. In this way, the sensori-motor behavior
of the individual participants spontaneously coheres to higher scalar patterns
which have to do with, for example, relational dynamics existing on quite
diVerent temporal scales. Sensori-motor dynamics such as vibrating vocal
chords are biological initiating conditions that self-organize into patterned
behavior at the level of the particular conversational event. In turn, the spon-
taneous emergence of such patterns is entrained to still higher level patterns
such as, for example, the historical-biographical unfolding of the social rela-
tions among the participants to any given conversational event. The relational
dynamics of this level provide boundary conditions which regulate and en-
train the dynamics of the lower levels to its patterns. In the examples analyzed
by Cowley (1998), the higher scalar level of the social group to which indi-
vidual dynamics are entrained is, in the Wrst instance, that of the family,
comprising the two parents and their daughter, which is a social unit having
its own dynamical patterns characteristic of that level and their history. In
postulating such patterns and their history on their particular temporal scale
— months and years as compared to centiseconds! — the analyst is, in eVect,
postulating an entity — a traditional one in this case — which has the possibil-
ity of becoming part of some theoretical discourse whereby diVerent scalar
levels come clearly into view. In this case, the focal level is that of the conversa-
tional event. The scalar hierarchy of relevant levels with reference to this focal
level can be described as a three-level system, each with its own processes and
entities unique to that level. Thus:
N+1: relational dynamics of family as social group and the history of its
evolving patterns
N: the conversational event comprising a number of individual partici-
pants whose behaviors cohere to the higher-scalar patterns of the overall
event
N–1: the sensori-motor dynamics of individual organisms as biological
initiating conditions
The hierarchy of levels featured here illustrates how language-in-use cannot

be explained in terms of a single, scale-homogeneous level of semiotic or other
relations. Instead, it is a cross-scalar, semiotic phenomenon which is able to
link potentially very many scales, ranging from the organismic to a diversity of
space-time scales extending ‘up’ to the history of an entire culture (Lemke
2000). This aspect of language thus draws attention to the principle of the
scalar heterogeneity of language which I alluded to in the title of this chapter.
For this reason, it is perfectly legitimate contra Gumperz (This volume: 121) to
claim that language and other semiotic modalities construe experience (see
also Halliday 2000b:223). This is so in the sense that a language exists on
space-time scales that go beyond the “shared inferencing of human actors”,
seen as “the locus of interpretive assessment” (Gumperz, this volume: 121).

Meaning-making does not occur solely on the here-now scale in which social
actors’ interpretive assessments take place; it is also a constitutive part of
processes on much larger spatial and temporal scales right up to long-term
historical processes of the evolution of a meaning system. Meaning-making
implicates complex cross-scalar semiotic and material relations operating
simultaneously over very many scales rather than being limited to the locally-
interpreted intentions and inferences of social actors interacting with other
social actors. Scalar heterogeneity is therefore intrinsic to the nature and
functioning of language (Lemke 2000; Thibault 2000). For this reason, scalar
heterogeneity must be an essential component in the theorization of language.
4. Contextualization cues as indexical signs
Gumperz, as we have seen above, makes a distinction between context-free

symbolic signs and context-dependent indexical signs of which contextua-
lization cues are a class (Prevignano and di Luzio, this volume: 10). The
former have propositional content; the latter do not. Contextualization cues,
Gumperz argues, “serve to retrieve the contextual presuppositions conversa-
tionalists rely on in making sense of what they see and hear in interactive
encounters” (Prevignano and di Luzio, this volume: 9). These function on the
basis of “direct conventional associations between signs and context, estab-
lished or transmitted through previous communicative experience” (Previg-
nano and di Luzio, this volume: 10). Such conventional indexical associations
arise as a consequence of the ways in which initially random associations
between indexical sign and what is seen or heard in the perceptual purview of
the interactants emerge from the micro-level dynamics of a given here-now
association so as to take on more global conventional associations on the
macro-level of the social group.
Prosodies and rhythm are grounded in the bio-physical characteristics
and potentialities of the human body. The characteristics of the body allow the
individual to perform a certain range of bodily behaviors. That is, the ana-
tomical and physiological structures and processes of the body qua biological
organism impose their own lower-level constraints on what the body can do
relative to its environment. However, the conventional indexical associations
referred to by Gumperz refer to collective patterns which emerge as a result of
the ways in which the bodily potential of the organism is entrained to larger-
scale patterns of social interaction. In other words, a large number of degrees
of freedom of our bodily dynamics on level N–1 of the three-level hierarchy
described in Section 3 is entrained to a more restricted set of possibilities in
virtue of the conventional indexical associations on level N through our
attempts to reciprocally adapt to the behavior of others in our interactive
encounters with them on level N+1. In this way, we see how the conventional
indexical associations discussed by Gumperz result from our ongoing eVorts
to adapt our own bodily dynamics to the bodily dynamics of other individuals
with whom we engage in interactive, goal-directed or purposeful activity.
Such mutually modulated body dynamics constitute emergent collective pat-
terns which, in turn, exert their own eVects and constraints on the activities of
individuals.
Gumperz describes such indexical associations as “context-dependent”
because their contextual eVects arise in immediate situations in the here-now
of a given interaction. This fact is evidenced by the ways in which the given
(indexical) sign is dependent on speciWc sensori-motor cues for its interpreta-
tion. Indexicals of the sort discussed by Gumperz under the rubric of “con-
textualization cues” allow the individual to interact with other individuals on,
roughly speaking, the space-time scale of their shared perceptual purview. In
this case, the principle of semiotic mediation at work here means that the
here-now body dynamics — e.g. the particular vocal or manual-brachial
gesture — which take place on the scale of seconds or fractions of seconds in
the real-time of our perceptual-motor activity, indexes contextual associa-
tions which are commensurate with the scale on which the gesture qua indexi-
cal sign itself is articulated. This is the scale on which perceptual-motor
activity and its environmental eVects is both enacted and observed through
the perceptual systems of the body in interaction with its ecosocial environ-
ment. In other words, semiotic mediation by pure indexicals already entails
scales beyond the organismic one per se at the same time that it is constrained
by higher-scalar constraints in the form of the conventional associations that
have emerged between indexical signs and their respective contextual values.
Figure 1 models the complex, non-linear, inter-level constraints that operate
here across a number of diVerent scalar levels, ranging from the organismic
scale of the human body to the higher-scalar, socio-cultural conventions
involved.
Now, the fact that contextualization cues are conventional associations
emergence of conventional associations
entraining of bodily dynamics to large-scale situation-specific value

or content patterns of human interaction
Index
bio-physical dynamics of human body expression
constraints on bodily dynamics
Figure 1. The entraining of body dynamics qua conventional indexical association to

higher-scalar ecosocial dynamics.
already implies that they have a history and that this history has both indi-
vidual and social dimensions. That is, a given use of a conventional indexical
association implies scales of dynamical complexity in the given ecosocial
system which go beyond the momentary use of a particular contextualization
cue in some context. How did the user come to use the particular con-
textualization cue? In relation to which social practices in the ecosocial sys-
tem? How did the given cue come to be part of the particular user’s repertoire
of meaning-making resources? In relation to which social networks and their
characteristic bodily dispositions (cf. habitus)? Contextualization cues such as
rhythm and prosodies have their basis in the neurophysiological substrate and
perceptual-motor activity of the organism, yet their eVects extend beyond the
organismic scale of the individual and into the ecosocial environment of other
social agents for whom they are potentially meaningful. The questions posed
here point to the fact that contextualization cues qua indexical signs are
embedded in at least two interpenetrating scales which have a history.
First, there is the personal-biographical history of the individual’s trajec-
tory through speciWc social networks. Secondly, there is the history of the
ecosocial system within which individual trajectories are in turn embedded.
Contextualization cues can only have their meaning in relation to both em-
bodied individual memory and the accumulated history of the meaning-
making practices of a given culture or some part of this. Indexical signs
mediate bodily processes so as to amplify their eVects on the ecosocial envi-
ronment, including other individuals, beyond the scale of the micro-level
processes which constitute their bio-physical substrate on the organismic
level. Indexicals integrate phenomena on the basis of the perceptual-motor

access of interactants such that the latter perceive the index as being contigu-
ous with a given contextual object or value. In this way, the use of a particular
prosody or rhythm is construed as being contiguous with an inferred commu-
nicative intent in the sense that the situated perception of the cue provides
access to the inferred intention (see Prevignano and di Luzio, this volume: 12–
14 for examples). The indexical inferring of such communicative intentions is
necessarily situated and on-line, so to speak. In other words, the relevant
space-time scales imply scales of dynamical complexity which are roughly
commensurate with the space-time of the unfolding meaning-making activ-
ity. While intentions cannot be directly perceivable, they are, however,
indexically invoked by on-line contextualization cues, based on perceptual-
motor access to the relevant domain. This means that the inferred intention is
an indexically-invoked contextual value — not a material object — which is
integrated with a perceivable cue through processes of semiotic mediation.
The fact that contextualization cues “channel the interpretative process”
(Prevignano and di Luzio, this volume: 10) so as to constrain the possible
meanings of an utterance demonstrates this last point. That is, the given
contextualization cue is selectively co-contextualized in relation to some other
features of the wider situation which the speciWc co-contextualization helps to
create or enact. In this sense, situated inferences as to what a speaker ‘intends
to mean’ in a given context are not external to the situation, but are a constitu-
tive part of the meanings which constitute and deWne that situation. Thus, the
use of, say, a particular prosody, rhythm or voice quality, in order to perform
the constraining function mentioned above, implies a number of space-time
scales, all of which are simultaneously and seamlessly interwoven in the real-
time unfolding of the communicative event. There is, for example, the scale of
the activity which is being performed. Furthermore, there is the scale of the
history of the personal relationship between the participants, which may vary
from the contingent and Xeeting to one spanning an entire life-time. There is
also the scale of the social networks in which particular prosodies and so on
typically redound with other semiotic features in the formation of a given
social situation-type. This last point brings me to a Wnal consideration of the
relations between semiotic forms and the discursive practices in which the
former are embedded.
5. Linguistic practice and discursive practice
According to Gumperz:
Sociolinguistic explanation, if it is to be relevant to today’s concerns, cannot
implicitly accept traditional categories of language, culture and society. I believe
that interaction at the level of discursive or communicative practices […] must
be seen as separate from either linguistic or socio-cultural processes. It is consti-
tuted by the interplay of linguistic, social and cultural/ideological forces and
governed or constrained by partly universal and partly local-speciWc organiza-
tional principles.
(Prevignano and di Luzio, this volume: 9–10)
This formulation of discursive practices, on the one hand, and linguistic and
socio-cultural practices, on the other, suggests that socio-cultural groups and
their associated practices have a prior existence with respect to the ways in
which they are constituted, construed and evaluated through language and
other semiotic activities, material processes, and related objects and artifacts.
In my view, any such separation is an analytical, rather than a constitutive, one
(Thibault 1991:229–240). Social formations are neither logically nor onto-
logically prior to the social semiotic and material activities in and through
which meanings are made. Instead, they are constituted by these activities.
Take Gumperz’s example of the Afro-American speaker’s use of the utterance
Ahma git me a gig (Prevignano and di Luzio, this volume: 19–20).
As Gumperz’s analysis shows, the lexicogrammatical and phonological
features which co-pattern in this utterance instantiate semantic patterns
which are typical of register-speciWc semantic relations that occur in the
speech of the Afro-American speakers referred to. Such patterns are charac-
teristic of the discourse community to which these speakers belong. More-
over, it is not simply a question of the speaker talking about a particular topic
in a way which diVers from the ‘academic norm’ of standard American En-
glish referred to by Gumperz. Rather, the speaker activates a small fragment of
a much wider intertextual thematic formation which is recognizable and
interpretable by members of the discourse community in question. It is a
thematic formation which is also typically associated with a particular evalua-
tive orientation or social viewpoint insofar as it implicates an evaluative stance
both towards his interlocutors as well as towards the thematic content of his
utterance. For this reason, the utterance can be seen as the activation of what
Bakhtin (1981) called a particular “social voice”, i.e. a particular social view-
point on some topic. For Gumperz, the particular co-patterning of lexico-

grammatical and phonological selections in the utterance functions indexi-
cally to convey the speaker’s message by the use of “metapragmatic strategies”
(Prevignano and di Luzio, this volume: 19) which index the speaker’s iden-
tiWcation with Afro-American social groups who talk in this way. Thus, the
utterance is an index of such sociolinguistic group membership or alignment
with the members of this group.
Nevertheless, it is clear that the choice of this particular utterance does
more than index group aYliation. It also realizes a social voice, in Bakhtin’s
sense, within a wider system of potentially alternative, conXicting, aligned,
and so on, evaluations and viewpoints concerning various topics. My point is
that the index of group membership, the theme or topic of the discourse, and
the evaluative stance which is invoked are not neatly separable categories: they
all come together and are variously realized by the linguistic choices made by
the speaker in the context in which he spoke. For this reason, the linguistic
processes analyzed by Gumperz, presumably along with other bodily features
associated by the given social habitus of the speaker, are said to be constitutive
of the given social formation on that particular occasion of talk. That is why, in
my Wrst contribution to this volume (Thibault, this volume: 50), I argued for
the three-way view of meaning-making practices articulated there as a way of
showing how particular associations of the evaluative, thematic, and other
meanings realized in linguistic and other semiotic forms are created in dis-
course at the same time that they directly implicate the higher-scalar social
formations relevant to their interpretation. It is the use of small-scale lexico-
grammatical patterns in potentially very many local contexts that leads to the
emergence of higher-scalar intertextual formations and their associated prac-
tices. A given usage in some local context can therefore invoke the entire
formation of which it is a mere fragment. By the same token, the higher-scalar
intertextual formation, which exists on the space-time scale of an entire com-
munity and its history, semiotically constrains speciWc, fragmentary instances
of the whole formation. Without some such system of the complex, multilevel
and non-linear constraints operating across all scalar levels from the most
global to the most local, it is diYcult to see how social meaning-making
practices shape social and individual life without having recourse to out-
moded models based on mechanistic cause-and-eVect relations. This further
suggests that higher scalar levels such as the intertextual formations men-
tioned here have their own ontological reality. Consequently, they cannot be
reduced to or theorized in terms of the lower level constituent units and

processes per se from which they emerge and which they in turn constrain.
That is one reason why I proposed in Thibault (this volume) the three interre-
lated perspectives on our meaning-making practices as a way of showing how
higher-scalar relations and processes enact their own constraints on the selec-
tions made in speciWc texts.
Gumperz’s claim that contextualization cues allow interactants to infer
intended meanings tends, moreover, to see meaning as internal cognitive
processes which originate from within the individual organism. Thus, con-
textualization cues such as prosody and rhythm are indexes which provide
access, by means of inferential processes, to the internal cognitive states of
individuals in interaction. This suggests that contextualization cues can be
‘read oV’ individual behavior as indexes which (indirectly) point to or are
contiguous with the postulated internal cognitive states of individuals. Fur-
thermore, there are macro-social interpretative conventions which allow con-
trasts in prosody and rhythm and so on to be appropriately ‘assessed’ and
‘interpreted’ as to their intended meaning in a given context (see Prevignano
and di Luzio, this volume: 14).
Gumperz further suggests that there are context-free signs which he
equates with symbolic, rather than indexical, modes of sign-making. In my
view, symbolic signs, as seen from the instantial perspective of their deploy-
ment in discourse, are no less context-dependent than are indexical signs.
Symbolic signs construe phenomena of experience — both physical-material
and imaginary — as instances of the semiotic category realized by the sign.
Furthermore, a symbolic system of possibilities such as language is formatted
in regular lexicogrammatical patterns which can allow for oV-line reXection
on the symbolic signs themselves as objects of knowledge in their own right.
Lexicogrammar embodies both indexical and symbolic functions. The
diVerence between the indexical and symbolic modes lies in the way in which
symbolic modes of semiosis make possible the maximal intersection of
diVerent scales such that a particular use of a symbolic sign in some here-now
context may implicate many diverse and heterogeneous spatial and temporal
scales, all of which seamlessly interact in a given use of a particular symbolic
sign in some context. Rather than being “context-free”, symbolic signs func-
tion to integrate potentially many diVerent contexts on varying space-time
scales (Thibault 2000), such that the texts and discourses which we make out
of our symbolic meaning-making resources are relatively small-scale units
which, however, directly implicate larger-scale social networks, larger com-

munity-scale systems of intertextual meaning relations and their histories,
larger-scale spatial extension beyond the physical-material reality of the tex-
tual artifact or the speciWc occasion of discourse in the here-now.
Symbolic semiosis does not necessarily entail that interactants have per-
ceptual-motor (indexical) access to the phenomena of experience which are
construed through the experiential resources of lexicogrammar. Symbolic
meaning-making resources allow for the integration of many diVerent space-
time scales which are not directly accessible to the interactants, but which can
be opened up and made accessible so that our relations to other scales beyond
the here-now can be both modeled and made relevant to the here-now scale of
our moment-by-moment interactions with our ecosocial environment.
Note
1. I wish to thank Carlo Prevignano for his valuable comments on an earlier version of this
chapter.
References
Bakhtin, Mikhail. 1981. “Discourse in the novel”. In The Dialogic Imagination: Four Essays
by M.M. Bakhtin, C. Emerson and M. Holquist (trans.) M. Holquist (ed.), 259–422.
Austin and London: University of Texas Press.
Bateson, Gregory. 1973 [1972]. “Problems in cetacean and other mammalian communica-
tion”. In Steps to an Ecology of Mind, 334–348. London and New York: Granada.
Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge, Mass.: MIT Press.
Cowley, Stephen J. 1998. “Of timing, turn-taking, and conversations”. Journal of
Psycholinguistic Research 27 (5):541–571.
Firth, J.R. 1957 [1948]. “Sounds and prosodies”. In Papers in Linguistics 1934–1951, 121–
138. London and New York: Oxford University Press.
Gumperz, John J. “Response essay”. This volume.
Halliday, M.A.K. 1979. “Modes of meaning and modes of expression: Types of grammati-
cal structure, and their determination by diVerent semantic functions”. In Function
and Context in Linguistic Analysis: A Festschrift for William Haas, D.J. Allerton, E.
Carney and D. Holdcroft (eds.), 57–79. Cambridge and London: Cambridge Univer-
sity Press.
Halliday, M.A.K. 1992. “A systemic interpretation of Peking syllable Wnals”. In Studies in
Systemic Phonology, P. Tench (ed.), 98–121. London and New York: Pinter.
Halliday, M.A.K. 2000a. “Phonology past and present: A personal retrospect”. Folia
Linguistica XXXIV (1–2):101–111.
Halliday, M.A.K. 2000b. “Grammar and daily life: Concurrence and complementarity”. In
Functional Approaches to Language, Culture and Cognition, D.G. Lockwood,
P.H. Fries, and J.E. Copeland (eds.), 221–237. Amsterdam and Philadelphia: John
Benjamins.
Lemke, Jay L. 1999. “Opening up closure: Semiotics across scales”. Paper presented at the
conference, Closure: Emergent Organizations and their Dynamics, University of
Ghent, Belgium, May 1999.
Lemke, Jay L. 2000. “Material sign processes and emergent ecosocial organization”. In
Downward Causation: Minds, Bodies and Matter, P. Bøgh Andersen, C. Emmeche, N.
Ole Finnemann, P. Voetmann Christiansen (eds.), 181–213. Aarhus: Aarhus Univer-
sity Press.
Levinson, Stephen C. “Contextualizing ‘contextualization cues’”. This volume
Pike, Kenneth L. 1967. Language in Relation to a UniWed Theory of the Structure of Human
Behavior. Second, revised edition. The Hague and Paris: Mouton.
volume.
Thibault, Paul J. 1991. Social Semiotics as Praxis. Text, Social Meaning Making, and
Nabokov’s Ada. Minneapolis and Oxford: University of Minnesota Press.
Thibault, Paul J. 2000. “The dialogical integration of the brain in social semiosis: Edelman
and the case for downward causation”. Mind, Culture, and Activity 7 (4):291–311.
Thibault, Paul J. Forthcoming. The Signifying Body: An Ecosocial Semiotic Theory of
Situated Meaning-Making, Embodiment, and Consciousness.
Thibault, Paul J. and Van Leeuwen, Theo. 1996. “Grammar, society, and the speech act:
Renewing the connections”. Journal of Pragmatics 25:561–585.
Chapter 10
Continuing the discussion with

John J. Gumperz
Carlo L. Prevignano and Paul J. Thibault
Carlo L. Prevignano: In the conclusion to your Response essay, you mention a

case study of yours, Communicative Competence as Communicative Practice.
For Gumperz in 2002, what are communicative and interactional com-
petences, and what are their relationships with practice? What solution do you
propose today to the attempt to capture “regularities of discourse practice
without reifying them in terms of a priori rules” (Gumperz, this volume: 122)?
John J. Gumperz: The case study, Communicative Competence as Communi-
cative Practice, is still unWnished. Its title alludes to the theoretical notion Wrst
introduced by Dell Hymes in the nineteen seventies in connection with his
argument against Chomsky’s narrowly-deWned “linguistic competence”.
Hymes’ claim was that there are additional forms of systematic knowledge
apart from grammatical rules, and that these also enter into the production
and comprehension of talk. He concludes that we need a broader, more
inclusive notion to account for situated language use along with linguistic
structure.
In the paper, I analyze an informal, student coVee shop conversation to
argue that communicative competence cannot be described in terms of
“sociolinguistic rules” abstracted from everyday talk and patterned on those
used in grammatical analysis, as is currently done in educational psycholo-
gists’ tests of English as second language learners’ speaking abilities. Commu-
nicative competence is more directly reXected in the way participants use
language to respond to others in negotiating shared understandings in natu-
rally-organized discursive encounters. Based on sample passages from the
transcript, my analysis seeks to illustrate the sort of things participants have to
know to engage in and maintain discursive coherence in interactive encoun-
ters. I argue that educators constructing tests of communicative competence

might concentrate on Wnding ways of assessing learners’ situated interpretive
ability. The resulting tests would not just ‘measure competence’, they might
also yield normally unavailable insights into how such communicative diY-
culties can aVect the course of an interaction that might improve teaching
programs.
For those who are interested, a second paper, Sharing common ground,
dealing with the same conversational exchange, makes similar arguments.
This paper is published in a Festschrift for Werner Kallmeyer (Gumperz
2002).
I don’t remember where I used the term “interactional competence”. Most
probably I intended to refer to knowledge of the kind alluded to in the two
above-mentioned papers. But let me clarify my basic assumptions a bit more.
I believe that if we want to determine how speakers understand each other, we
must account for all signs that demonstrably aVect the course of an encounter.
I use “communicative practice” (CP) as a general term to cover the relevant
verbal and non-verbal signaling processes. Strictly speaking, “discursive prac-
tice” is a subcategory referring to the discourse-level elements of CP. But the
two terms are sometimes used interchangeably. Along with Silverstein (1992,
1993) and Hanks (1996), whose terminology I build on here, I see communi-
cative practice as constituted by three elements. First, language, roughly de-
scribable as a Saussurian system of abstract (symbolic) categories and rules for
their combinations. Second, contextualized communicative activity includ-
ing: (a) symbolic or lexicogrammatical signs; (b) indexical signs such as
prosody, rhythm, code- or style-switching, formulaic expressions and other
cues; (c) nonverbal signs such as gesture, posture, facial expression, clothing
etc.. Third: (a) ideology, in as much as it motivates actors’ meta-communica-
tive evaluations, based, among other things, on cultural beliefs and values; (b)
background knowledge relevant to what transpires in an encounter. Like
Thibault, I claim that all three elements are always simultaneously involved in
the production and interpretation of talk. That is, at the level of CP, talk is
grammatical, contextualized and ideologized. This tripartite division has
some similarity to Pike’s three dimensions discussed by Thibault in his second
paper in this volume. But as I will point out below, my view of the semiotic
processes involved is signiWcantly distinct from his. The three elements can of
course be, and have for a long time regularly been, abstracted from talk for
purposes of in-depth analysis and thus treated as ‘analytically distinct’. But
this means we must see them as abstractions and cannot claim that any one
element or constituent thereof can by itself account for the phenomenological
facts of communicative practice.
Turning to the quote at the end of your question, in using that expression,
I meant to argue against scholars who resort to extra-communicatively
deWned, largely lay categories, like class, gender, ethnicity, professional iden-
tity, community or group membership and the like, to explain or naturalize
what happens in situated interactive exchanges, without demonstrating how
the relationship works. As the preceding paragraph implies, some forms of
social/cultural/ideological knowledge always organize our actions and de-
monstrably aVect CP interpretive assessments. But this is not equivalent to
claiming that participants’ membership in one or another human collectivity
as such is directly reXected in talk. What we are talking about here is how
interpretive presuppositions, ideologically associated with such a collectivity,
aVect interpretation. The nature and workings of such presuppositions must
be investigated through the analysis of context-bound verbal exchanges.
As to “regularities” of communication practice, I believe that these should
ultimately be derived from or related to in-depth analyses of situated encoun-
ters in a variety of settings. Because of the way interaction works, the usual
questionnaire studies are not satisfactory here. I made some informal com-
parisons between South Asian and English interviewing practices in a paper of
mine (Gumperz 1996), but these are observations that might inform ethno-
graphic studies. I don’t know of any systematic research along these lines.
C.L.P.: At the end of your Response essay, you indicate a “distinction [analy-
tically of great importance, I agree] between explicitly metacommunicative
[and consciously manipulable] signs and indirect indexicals [which are pro-
duced below the level of consciousness] such as contextualization cues”. Do
you see today any other diVerence between the former and Gumperzian cues?
J.J.G.: On second thought, it might be more fruitful here to speak of a distinc-
tion between overtly-lexicalized indexicals, on the one hand, and non-
lexicalized or indirect indexicals (including, among others, prosody, rhythm,
tempo, certain phonetic or grammatical variables, along with code- and style-
switching), on the other. That would avoid value-laden terms. Linguistic
anthropologists, following Silverstein (1992, 1993), sometimes use the term
“pure indexicals” to refer to the latter category. That is to say, they are pure in
the sense that they basically function relationally or inter-discursively. Unlike
lexicalized indexicals, they have no phonetic substance of their own and

hence, are not meaningful when isolated from the stream of talk. In other
words, we cannot, for example, treat stress, rhythm, intonation contours as
having meanings of their own.
In a way, indirect indexicals are a bit like Benjamin Lee Whorf’s covert
categories. They readily lend themselves to ideological re-interpretation or
displacement, such that individual speakers become invested with the atti-
tudes, abilities (or presumed lack thereof), stereotypically attributed to the
group or occupational category indexed by their ways of speaking. Such
ideological displacements abound in everyday life, particularly in today’s
increasingly diverse urban societies. Empirical studies on intercultural and
interethnic communication in classroom teaching, performance or ability
assessments, counselling sessions, courtroom testimonies and so on, docu-
ment the eVect they can have on individuals’ life chances (see, for example,
Gumperz 1982; Jacquemet 1996; Young 1994).
From a more general theoretical perspective, treating prosody, rhythm
and style shifts as information-bearing indexicals calls attention to the perva-
sive importance of indexicality in everyday communication and suggests new,
replicable ways of recovering the unverbalized assumptions that underlie
what is actually said.
Last but not least. Since basic indirect indexical practices are acquired very
early in a child’s life, they are diYcult to unlearn. Even speakers who speak a
second language well often show traces that remind us of possible historical
origins of these practices (Gumperz 1996:390). Relatively little work has been
done exploring the import of this fact. Additional comparative studies might
show that more systematic comparative studies of contextualization-based
inference lend a historical dimension to Interactional Sociolinguistics (IS)
that other related traditions of discourse analysis lack.
C.L.P: For those who could not attend the IPrA Congress in Budapest, and/or
the Special Event Panel dedicated to you (July 10, 2000), could you sum up
your Wnal ReXections?
J.J.G.: Many of the papers dealt with the authors’ own work. After the panel
presentations, there was time for only a few minutes of oral comments. I have
no record of my comments.
Paul J.Thibault: In the Discussion, you state your belief that “interaction at the
level of discursive or communicative practices […] must be seen as separate
from either linguistic or socio-cultural processes” (Prevignano and di Luzio,

this volume: 9). Could you clarify the nature of the relationship, as you
envisage it, between discursive practices, on the one hand, and linguistic and
socio-cultural practices, on the other? Do socio-cultural groups and their
associated practices have a prior existence with respect to the ways in which
they are constituted, construed and evaluated through language and other
semiotic activities, material processes, and related objects and artifacts? In
your view, is the distinction an analytical or a constitutive one?
J.J.G.: As it stands, the statement you refer to is somewhat confusing. What I
meant to suggest is that Wndings based on CP analysis are distinct from
Wndings based on either linguistic or socio-cultural analysis, since these forms
of analysis build on diVerent kinds of source data and rely on diVerent modes
of validation. I therefore suggested that the three dimensions are not compa-
rable.
As I implied in the answer to Prevignano’s second question, I see linguistic
and socio-cultural knowledge as analytically distinct, yet constitutive ele-
ments of communicative practice. As for the second part of your question, let
me repeat: what is of primary importance from an IS perspective, is what we
believe about such groups and how these beliefs aVect interpretation, not
whether or not the groups have a “prior existence”.
P.J.T.: In your Response essay, you object to the claim (that I also make in my
second contribution to this volume (Thibault, this volume: 138)) that lan-
guage and other semiotic modalities construe experience. In your view, the
agency is located at the level of the individual social actors who engage in
“shared inferencing”. But given that language and other semiotic modalities
exist on space-time scales that go beyond the “shared inferencing of human
actors”, seen as “the locus of interpretive assessment” (Gumperz, this volume:
121), don’t you think that larger space-time scales are also implicated and that
agency is distributed across these, rather than being located in social actors per
se? How do you see the relationship between the space-time scale on which
actors engage in shared inferencing and the larger spatial and temporal scales
involved?
J.J.G.: What I intended to convey in the Response essay was that from an IS
perspective, participants construe experience via interpretive assessments.
This does not mean that I am taking a radical constructionist or post-
modernist position. I argue that what you call knowledge [of linguistic struc-
tures] existing on “larger spatial and temporal scales” has its origin in con-
ventions that arise in interaction and over time have become grammaticalized.
These may then be brought into interaction by participants as part of the
interpretation process. Like other grammatical knowledge, these conventions
are subject to preservation, reaYrmation or change through what happens in
later communicative practice.
I do of course agree that we need to make a distinction between what
I would call “interactional time” or, as you put it, “the space-time scale” of the
interactional exchange and “larger spatial and temporal scales”. But what I
have to say does not really contradict that.
Your remarks concerning my objection to the claim that “language and
other semiotic modalities construe experience” go to the heart of the diVer-
ences between your and the IS approach. Your systemic perspective leads
you to privilege formal semiotic structure as a source of insight into the
nature of linguistic processes. IS, on the other hand, argues that we have no
access to experiential reality, except as it is Wltered through our own or some
other human interpreter’s mind, so that we must look to communicative
practice. This is why I emphasize shared inferencing. The very fact that
interactants maintain conversational involvement over a string of move-
countermove-rejoinder sequences constitutes empirical proof that they
share the presupposed grammatical and indexical knowledge, since the in-
ferential process constitutes the cognitive space where grammatical and in-
dexical knowledge meet. But this does not mean that “language and other
semiotic modalities” existing “on space-time scales” that go beyond the
“shared inferencing of human actors” are not relevant for IS. What IS sug-
gests is that we should not assume there is a one-to-one relationship between
the linguist’s bodies of abstract rules and regularities and what goes into
shared inferencing. The relationship between the two needs to be investi-
gated in specially-designed studies. Until this is done, the two approaches
will remain largely incommensurable.
C.L.P.: Many years ago, you proposed the interaction-oriented notions of
“linguistic repertoire”, “linguistic community”, and later, “social network”.
How do you see these today?
J.J.G.: The notion of “speech community” or “linguistic community” (the two
terms were often used interchangeably, but see my Wnal remarks), has always
been rather loosely employed to refer to the social space in which language is
spoken and with respect to which it is evaluated. In what follows, I will

concentrate on my own use of the concept.
I Wrst became aware of the deWnitional problems that can arise in the
course of my dissertation research on the Swabian dialect of Washtenaw
County, Michigan. A number of results emerged: (a) a survey of local second
and third generation dialect users showed that this relatively small popula-
tion, normally seen by outsiders as a single community, was divisible into
three groups or subgroups, each with its own distinct speech variety; (b) a
comparison with the dialect situation in the country of origin showed that
individual speakers’ language use did not directly reXect the ancestral dia-
lect; (c) local speech varieties had undergone signiWcant change in the two or
three generations of residence in the United States; and (d) the main factor
determining current language use was membership in one or another of the
three local Lutheran church groups which also served as the main centers of
local social life.
My postdoctoral research in India followed the then generally-accepted
practice of taking the village community as the initial universe of description
and analysis. But when I looked at the communicative bases of community
membership, it soon became evident that the dialect isoglosses I discovered
did not follow caste and community boundaries. Finally, in the course of my
stay in India, I became aware of the multilingual communities in South and
South East Asia, such as those studied in Edmund Leach’s Hpalang in High-
land Burma (in Leach 1954), where four grammatically distinct and mutually
unintelligible languages are spoken in what, on socio-political grounds, must
count as a single community. Based on such data, I argued that while a speech
community, if it is to function, must constitute a communicative whole, we
cannot tie the deWnition to one or more particular languages or varie-
ties thereof.
The 1968 deWnition of the speech community (“any human aggregate
characterized by regular and frequent interaction by means of a shared body
of verbal signs and set oV from similar aggregates by signiWcant diVerences in
language usage” (Gumperz 1968:381)), which was widely discussed at the
time, reXects this position.
I adopted the term “linguistic repertoire” to account for intra-community
diversity. A repertoire was said to consist of a set of speech varieties which
constitute the totality of the community’s linguistic resources. The assump-
tion was that members of a community choose among the varieties of the
repertoire, depending on what is appropriate in a particular context of speak-

ing. When speakers regularly alternate among varieties of a repertoire, they set
in motion processes of linguistic convergence which gives the local group a
linguistic identity of its own that sets it oV from the surroundings. The
Kupwar study referred to in the Response essay describes an extreme case of
change where local varieties of genetically-unrelated languages come to con-
verge through regular and frequent contact over several hundred years.
In hindsight, the above considerations can be seen as initial steps towards
a socially-constituted sociolinguistics that accounts for shared systems of
verbal signs without, however, privileging speciWc languages or varieties
thereof. The exclusive focus on ongoing communication in the discussion so
far is also limiting in important respects. I became aware of this Wrst in the
work with J.-P. Blom in Northern Norway, where we noticed a linguistically
signiWcant distinction between those who saw themselves as members of the
“local team” of friends and neighbors, and those who, for professional rea-
sons, also maintained regular relations with outside members of the northern
Norwegian commercial elite. What we were talking about in this case were
networks of relationships, although at the time we did not overtly state this.
The network concept becomes central in the research in the Slovenian/Ger-
man border area (Discourse Strategies, Chapter 3), where I argued that Slove-
nian is disappearing because of restructuring of social networks caused by the
economic changes in the region. Had I focussed just on intra-community
relations, I could not have explained the changes that concerned me, since the
notion “speech community” denotes a more or less bounded group, or one of
a series of such entities. A network, by contrast, can and often does denote
relationships not necessarily limited to any one locality. The term “network” is
by now widely used in sociolinguistic research. But whereas most socio-
linguists tend to treat networks as associations of individuals, my own work,
as illustrated in the Introduction to section 4 of Rethinking Linguistic Relativ-
ity (Gumperz and Levinson (eds.) 1996), sees networks as constituted by
social relationships established over time, which are linguistically marked and
indexically associated with speciWc communicative goals which enter into the
evaluation of what is said.
A Wnal, serious limitation of my earlier work is its lack of attention to
linguistic ideology. In a way, this notion is implicit to the argument in the
introduction to Ferguson and Gumperz’s Linguistic Diversity in South Asia
(1960), where we drew a distinction between languages as “socially-deWned”
units — as we put it at the time — and speech varieties as “linguistically-

deWned” units. We cited a number of cases, where what I would now call
“linguistic ideology” led individuals to identify themselves as speakers of one
language, whereas grammatically their speech counts as a variety of another.
The workings of ideology are clearly brought out in Michael Silverstein’s
Monoglot ‘Standard’ in America: Standardization and Metaphors of Linguistic
Hegemony (1996). Silverstein also makes an important distinction between
speech communities “sharing a set of norms and regularities for interaction by
means of languages” and linguistic communities deWned as “a group of people
who, in their implicit sense of the regularities of linguistic usage, are united in
adherence to the idea that there exists a functionally diVerentiated norm for
using their ‘language’ [...], the inclusive range of which the best language users
are believed to have mastered in the appropriate way” (1996:285). These
deWnitions are the most useful I have seen to date, since they account for the
pervasive normativeness and linguistic hegemony characteristic of our cur-
rent nation states. But there is no space to discuss this here.
C.L.P.: A question concerning Gumperz as an analyst of interaction, with a
preamble. The idea of inference that is at the center of your interpretive
approach to human “interaction order” connects culturally speciWc “practical
reasoning” and presuppositions with interaction surface and cue processing.
Cross-inferencing and misinteractions are possible just because conversa-
tions are not simply exchanges of cues corresponding to behaviouristic pat-
terns of cue/response and response/cue. Several studies seem, on the contrary,
to persist in adopting an implicit behaviouristic approach to cue- and turn-
processing, while you have been proclaiming (at least since 1982) the relevance
of interactional and conversational “strategies” in your analyses, with a focus
on members’ or conversationalists’ interpretive, inferential procedures.
Starting from some Gumperzian cues, which I have reconsidered in rela-
tion to human interaction protocols and ascribed to the class of interaction
cues, I have tried to identify as “minims” (Prevignano, this volume) some
default rules governing the human use of activity-type cues in both output
(cue emission) and input (cue reception and interpretation). In my view of
cue processing, minims correspond to minimal tactical requirements for
Wnding economical human solutions to everyday interaction problems; these
solutions would be much more diYcult and time-consuming, not only with-
out cues, but also without a minim-governed use of them. What is at stake
here, as in the study of other human interactional tasks and abilities, is an idea
of interactional intelligence (representable as a cultural set of strategies, tac-
tics, heuristics, default rules, meta-rules, and so on), as has been anticipated
and practiced by your IS research over a long period of time. The construct of
interactional intelligence is necessary, in my opinion, in order to deal with the
study of human, interactionally intelligent conversationalists.
After this long premise, I come to a short question: could you, as a de facto
analyst of interactional intelligence in and between diVerent cultures, give
some suggestions for future research on it?
J.J.G.: Let me restate parts of the question’s preamble in my own terms,
expanding on what I understand the questioner to have said and on argu-
ments I made elsewhere in this volume.
A. As to interaction analysis. Along with other Linguistic Anthropolo-
gists, I assume that we, that is, specialists as well as lay persons, have no direct
access to physical, social or psychic reality, except as it is processed by some
interpreter’s mind. It is this processing, as reXected in the inferences partici-
pants draw on, on the basis of what they believe they see, hear, recall, or
otherwise perceive in speciWc sets of circumstance, that lies at the center of IS
analysis. So that the focus is no longer on word-to-world relationships, but on
the ‘how’ of interpretation, i.e. the procedures we and others rely on to assess
experience.
B. For analytical purposes, it is useful to treat the inferential process as
made up of several stages. In the Wrst of these stages — the most basic one in
the sense that it underlies further, higher-level inferencing — perceptions are
(among other things) chunked, categorized into event types and phrases and
assigned phonetic, prosodic, rhythmic values. This yields what, using a
modiWed version of Silverstein’s (1993:36–8) term, I call an “interactional
text”. As I use the term, the notion of interactional text makes explicit the
assumptions of lay persons and analysts about what they perceive, and at the
same time provides the indexical ground for further interpretations. Unlike
Saussurian analysis where etic constituents of talk are seen as part of the
physical world, an IS approach treats what the preamble calls “interaction
surface and cues” as components of interactional texts and thus as interac-
tively constructed.
C. As to “cross-inferencing and misinteractions” (I prefer the term “inter-
action breakdown” for the latter). My research shows that these are frequently
due to interpretive diVerences at the level of interactional text. The example

“Ahma git me a gig”, cited by Prevignano (this volume: 19–20), is a case in
point. When I Wrst looked at the example, I simply did not hear the singsong
intonation which later proved to be crucial for the interpretation. It was only
through interaction and direct questioning of my initial perceptions that I
became sensitized to the contouring and its communicative import. After that
I gradually became aware of its general semiotic import by transcribing and
analyzing other similar materials.
D. Turning now to the actual question about interactional intelligence
(personally, I prefer a more modest term like “interactional skills” or perhaps
“interactional competence”). You are right, my work raises the question of
what we have to know to signal active participation in interaction, and I
believe that in today’s increasingly diverse societies, where cultural bound-
aries are frequently not readily perceptible, this type of work is becoming
increasingly important. What I believe my work does is to suggest some initial
insights into the relevant issues and to propose theoretically-grounded em-
pirical methods of attack. But so far the issue has not been systematically
topicalized as a main object of research. Perhaps your term “interactional
intelligence” might help as an attention-raiser. In any case, we need more
comparative research contrasting how people signal participation in interac-
tions where communicative background is shared, with interactions where
sharing cannot be taken for granted, in societies around the world.
P.J.T.: How do you anticipate future developments of IS in the light of current
trends in the Weld? How do you see IS in relationship to other schools with
similar interests?
J.J.G.: Let me begin with the second part of the question. I have already
referred to the diVerences between IS and Conversational Analysis in earlier
parts of this volume. There are a number of schools that share the IS interest in
a linguistic analysis that can illuminate public issues in today’s societies. One
that has received a great deal of attention recently is Critical Discourse Analy-
sis (CDA), as reXected, for example, in the work of Norman Fairclough
(1989), Carmen Caldas-Coulthard and Malcolm Coulthard (1996), and Ruth
Wodak (1997). Like IS, CDA is grounded in talk and pays attention to sequen-
tial positioning of moves. But beyond that, CDA seeks to account for societal
factors like ideology, power and inequality, whether or not these are reXected
in talk. In a separate development, Linguistic Anthropologists in the U.S. have
also increasingly begun to turn to discourse analysis to document the role of

culture and social organization in human communication. An excellent ex-
ample here is Judith Irvine’s (1993) study of insulting practices among the
Wolof, which shows how locals use their knowledge of social structures to
manipulate roles in order to (among other things) dissociate a speaker from
responsibility for an insult. What distinguishes IS from these traditions is its
attempt to develop a general analytical frame capable of dealing with commu-
nicative situations of all kinds, which takes diversity of language and culture as
a given.
As to part one of the question, I have already pointed out the need for
more general treatments of conversational inference and its import for under-
standing our current, inherently diverse environments. There is also a need
for more systematic, interactively-based studies of the workings of linguistic
ideology in human interpretation. I am currently cooperating with Jenny
Cook-Gumperz on a paper on the ideology of standardization in modern
nation states and its eVect on educational policy.
References
Caldas-Coulthard, Carmen R. and Coulthard, Malcolm. 1996. Texts and Practices: Readings
in Critical Discourse Analysis. London and New York: Routledge.
Fairclough, Norman. 1989. Language and Power. [Language in social life series]. London:
Longman.
Ferguson, Charles and Gumperz, John J. (eds.). 1960. Linguistic Diversity in South Asia.
International Journal of American Linguistics 26 (3), special issue.
Gumperz, John J. 1968. “The speech community”. In International Encyclopedia of the
Social Sciences, J. Sills (ed.), 9:381–386. New York: Crowell Collier and Macmillan.
Gumperz, John J. 1992. “Contextualization and understanding”. In Rethinking Context:
In Rethinking Linguistic Relativity, J.J. Gumperz and S.C. Levinson (eds.), 374–406.
Gumperz, John J. 2002. “Sharing common ground”. In Soziale Welten und Kommuni-
kative Style: Festschrift für Werner Kallmeyer zum 60. Geburtstag, I. Keim and W.
Schütte (eds.), 47–56. Tübingen: Gunter Narr.
Gumperz, John J. “Response essay”. This volume.
Gumperz, John J. and Levinson, Stephen C. (eds.), 1996. Rethinking Linguistic Relativity.
Hanks, William F. 1996. Language and Communicative Practices [Critical Essays in An-
thropology Series]. Boulder, CO: Westview Press.
Irvine, Judith T. 1993. “Insult and responsibility:Verbal abuse in a Wolof village”. In
Responsibility and Evidence in Oral Discourse, J. Hill and J.T. Irvine (eds.), 105–134.
Jacquemet, Marco. 1996. Credibility in Court: Communicative Practices in the Camorra
Trials. Cambridge: Cambridge University Press.
Leach, Edmund R. 1954. Political Systems of Highland Burma. London: G. Bell and Sons.
Prevignano, Carlo L. “On Gumperz and the minims of interaction”. This volume.
volume.
enough?” In The Contextualization of Language, P. Auer and A. di Luzio, 55–76.
Silverstein, Michael. 1993. “Metapragmatic discourse and metapragmatic function”. In
ReXexive Language: Reported speech and Metapragmatics, J.A. Lucy (ed.), 33–58.
Silverstein, Michael. 1996. “Monoglot ‘standard’ in America: Standardization and meta-
phors of linguistic hegemony”. In The Matrix of Language: Contemporary Linguistic
Anthropology, D. Brenneis and R.K.S. Macaulay (eds.), 284–306. Boulder, CO: West-
view Press.
Thibault, Paul J, “Body dynamics, social meaning-making and scale heterogeneity: Re-
considering contextualization cues and language as mixed-mode semiosis”. This
volume.
Wodak, Ruth. 1997. Disorders of Discourse [Real Language Series]. London: Longman.
Young, Linda W.L. 1994. Cross-talk and Culture in Sino-American Communication. Cam-
Bio-bibliographical note
John J. Gumperz was born in 1922 in Germany, and emigrated to the United States in
1939. Currently Professor Emeritus of Anthropology at the University of California,
Berkeley, in 1954 he was awarded his Ph.D. in German Linguistics by the University
of Michigan, after Wrst completing a B.A. in Chemistry at the University of Ohio,
Cincinnati in 1947. His doctoral thesis, which was a study of the Swabian dialect of
third-generation farmers in Washtenaw County, Michigan, marked the beginning of
his interest in sociolinguistics. In this study, he formulated his early thinking on the
relations between social group membership and sociolinguistic processes. This was
the Wrst of a long series of research projects and Weld studies. On completion of his
doctorate, he undertook two years of Weldwork in India. He subsequently founded the
Hindi-Urdu program and was later Chairman of the Center for South and South East
Asia Studies (1968–71), both at Berkeley.
In 1956, Gumperz began collaborating with Charles A. Ferguson when both were
visiting faculty at Deccan College. The ensuing exchange of ideas on linguistic diver-
sity and language development led to their organizing a symposium at the 1958
annual meeting of the American Anthropological Association. The jointly published
volume, Linguistic Diversity in South Asia (1960), represents a landmark in the Weld.
Since 1966, Gumperz has been an active member of the Committee on Socio-
linguistics of the Social Science Research Council. This work led to the publication of
A Field Manual for Cross-Cultural Study of the Acquisition of Communicative Compe-
tence (1967), which has played an important role in the development of cross-cultural
research on sociolinguistic processes. He also collaborated with Dell Hymes, Susan
Ervin-Tripp, Edward T. Hall, Erving GoVman, Charles O. Frake, and William Labov
in the organization of symposia at the spring meeting of the Kroeber Anthropological
Society and the annual meeting of the American Anthropological Association. The
seminal volume, The Ethnography of Communication (1964), edited by Hymes and
Gumperz, was the product of these meetings.
Gumperz Wrst took up the position of Professor in Anthropology at Berkeley in
1965. In 1992, he was awarded a Dr. Hon. Causa by the University of Konstanz in
Germany. He was made Fellow of the American Academy of Arts and Sciences (1992),
the Max Planck Project Group in Cognitive Anthropology in Berlin (1991), Churchill
College, University of Cambridge (1987–88), as well as being a Senior Post-doctoral
Fellow at the National Science Foundation (1961–62). He was LSA Professor at the
Linguistic Institute of America, Georgetown University (1985). He has held Visiting
Professorships at the Universities of Bergen (1982) and Konstanz (1985).
Professor Gumperz’s major research interests lie in the areas of interactional
sociolinguistics, linguistic anthropology, social anthropology, ethnography of com-
munication, cross-cultural and interethnic communication, and bilingualism. His
major publications include Rethinking Linguistic Relativity, co-edited with Stephen
Levinson (Cambridge 1996); Engager la Conversation (Paris 1989); Sociolinguistique
164 Bio-bibliographical note
Interactionnelle: Une Approche Interprétative (Paris 1989); Discourse Strategies (Cam-

bridge 1982); Language and Social Identity, edited (Cambridge 1982); Directions in
Sociolinguistics, co-edited with Dell Hymes (New York 1972); Language in Social
Groups (Stanford 1971); Conversational Hindi-Urdu, with June Rumery (Berkeley
1963). He is also Editor of the series, Interactional Sociolinguistics, at Cambridge
University Press. His work in progress includes two book-length studies, Intercultural
Encounters and Language in Social Theory.
A bibliography of John J. Gumperz can be found at http://sunsite.berkeley.edu//
Anthro/gumperz/gumppub.html
Subject Index
accenting 91, 99 n. 14, n. 15 involvement 8, 9, 14, 100 n. 21, 122, 154

act/action/activity typologies/tokenologies communicative (or discursive) practices 8,
64-65 9, 11, 18, 41, 42, 94, 98, 112, 114,
activity 8, 13, 14, 15, 86, 93, 99, n. 2 120, 121, 143, 150, 151
activity type 13, 14, 15, 82 community 41, 54, 82, 109-110, 119, 120,
– analysis 72 121, 135-136, 143, 151, 154, 155
– application 63, 65 —see also speech communities
– approach 72 competence 3, 136
– cues 63 compounding 93, 95
– minims 63 computational linguistics ix, 79-80
– theory 72 – applied computational linguistics 83
Afro-American English 20 Computer-Supported Collaborative Work
ambiguity 17, 115 (CSCW) 80, 83 n. 2
analogic communication 129 context 3, 4, 6, 22, 42, 49, 86, 88, 90, 93, 98
contextual redundancy 49
background information 50, 53, 55 contextualization 4, 57
background knowledge 4 contextualization conventions 93
Bahktinian situation 94 contextualization cues vii, viii, 4, 5, 9, 10,
Black English 115 16, 32, 33, 34, 36, 38, 45, 58, 82, 88,
bodily dynamics 140 112, 113, 114, 127-129, 131, 133, 139-
boundary conditions 138 141, 145
– as pure indexicals 9, 14
central/peripheral information 36 – see also cues, Gumperzian
central/peripheral processing 68 contextualization strategies 93-94
clause grammar 52 contrast sets 56, 130
co-construction 97 convention 15-16, 26, 86-87, 93, 98, 154
code-switching 3, 31, 90, 108-109, 112, 114 conventional implicatures 34
cognitive saliency 35, 36 conversation analysis 8, 11, 14, 86, 94-95,
common ground 98 117-118
communication analysis vii, 66 conversational assessment 11, 19, 21, 118,
communication (or communicative) 121
ecology vii, 86, 89 conversational inferences, inferencing 10,
communication failure ix, 5, 89, 92-93 11, 12, 19, 86-88, 90, 93, 95-97,
communicative action 86, 88 100 n. 26, n. 27, 112, 117-118
communicative competence 93-94, 97, 100, conversational management 86, 90, 99 n. 3
n. 21, n. 25 conversational repair 17, 93, 95-96
communicative experience 87, 89, 93, 98 cooperation 93, 97, 100 n. 27
communicative intent 11, 86, 142 corpus linguistics 21
communicative (or conversational) corpus pragmatics 21, 83, 120
166 Subject index
correlational methods 7 illocutionary force 86

Croatian 106 implicatures 35, 86, 88
cross-examination 88-89, 94-97, 99 n. 2 indexes, referential 87
cue analysis 68ff. indexical practices, indirect 152
cue processing 68, 157 indexicality 57, 87, 97, 113, 152
cueing 68 indexicals (or indexical signs) 9, 10, 11, 45-
cues, Gumperzian 35, 67, 69 46, 49, 87-88, 95, 100 n. 20, 113, 121, 123,
culture 85, 86, 89 139-142, 145, 150, 151
cultural knowledge 89, 99 n. 9 indexicals, pure or indirect 88, 151, 152
inference 4, 86, 93, 142
deictic categories 48 interaction analysis vii
dialectal variability 1 interaction, human ix, x, 63, 69, 88, 94, 122,
dialectology 1 157
dialogue 19, 81 – as a multitrack space 69
– and reasoning 7, 19ff, 81 – an inferential model of 72
digital communication 129 interaction, robust 80
diversity x, 1, 85, 98 interaction order 8, 157
Dutch 106 interaction principle, semiotic 67-68
interactional competence 97, 100 n. 25
ecosocial environment x interactional intelligence 72, 158, 159
English 25, 48, 130, 131 interactional linguistics 70
entextualization 120 interactional sociolinguistics (IS) ix, x, 3, 6,
envisionments 13 31, 85-103, 118
ethnography 3, 89-90 interactional sociolinguistic analysis (IS
ethnomethodology 86 analysis) 86, 88, 90, 98-99,
117, 118-119
face-threatening act 95 —procedures of 11-12, 86, 88-97
frames 4, 10, 11, 13, 27, 37, 59 n. 6, 93 interactional texts 12, 90, 99 n. 12, 120, 158
framing 14, 27 interactive norms 92
French 106 interactive practices 97, 100 n. 25
interactive strategies 97-98
genre 3, 15, 28, 43, 44, 54, 56, 59 n. 6 intercultural communication ix, 5, 89-90,
German 106 92-93, 98
German Swabian dialect, see Swabian intertextual formation 144
Gricean analysis 86 intertextuality 53
Gumperz level (of interaction analysis) 67- intonation 4, 47, 58-59 n. 5, 131, 137
68 isochrony 96-97, 100 n. 24
Italian 15, 106
habitus 55, 141, 144
Head noun 48 job interview 88-94, 99 n. 2, n. 13, n. 14
heteroglossia, social 54
heteroglossic diversity 43 Kannada 2, 108, 109
Hindi 4, 100 n. 18
historical method 11 language as particle, wave, and field 132
Subject index 167
language shift 1, 86 nominal group 48, 130

language system 136 noncooperation 91, 93, 95, 97
language variation 1 nonlinear constraint 144
language variety 2 norms, social 2, 3
langue 3, 43 Norwegian, Standard 108
lexicogrammar 43, 44, 53, 131-133, 145 noun (see nominal group)
linguistic ideology 156-157, 160
linguistic repertoire 3, 110, 155-156 parole 3
participant observation 89, 99 n. 11
mapping 91 participation analysis 72
Marathi 2, 108 performance 3, 136
meaning phonology 131
– context-dependent 45 Piedmontese 25
– context-independent 45, 47, 51 pleading 92
– referential 47 postdictions 71
meaning-making practices postmodern condition 66
– indexical viii, 50, 51-53, 58
postmodernist conclusion 38, 65
– intertextual viii, 50, 53-57, 58
power asymmetry 5
– metadiscursive viii, 50, 56-58
power structure 5
meaning potential 48, 52
pragmatic creativity 21, 25-27
metacommunication 123
preference organization 14
metafunction 132
Principle of Alternation 133-134
metaphoric switching 112
prosody 44, 88, 90-91, 96-97, 100 n. 16, n.
metapragmatics 59 n. 6, 87, 99 n. 6
metapragmatic assessment 15 18, n. 24, 114, 131, 139, 141, 142, 145
metapragmatic function 100 n. 20 Punjabi 2
metapragmatic strategies 19, 20
metasystem 56-57 reconstruction, ethnographic ix
minimists/maximists viii, 63 recording, audio/video 90
minims vii, 63, 64, 122, 157 repair, see conversational repair
– input minims 63-64 rhetorical effectiveness 98
– interaction minims viii, ix, 63-64 rhetorical strategies 97
– output minims 63-64 rhythm 4, 44, 131, 139, 141, 145
– participation minims 64 rhythmic organization 90, 96-97
minims/maxims 122 Romansch 25
misunderstanding ix, 66, 89, 91-93, 98, 100
n. 19 scales, organizational 134, 137
mixed-mode semiosis x, 129, 133-134 scalar heterogeneity 139
move 12, 13, 14, 64, 65, 87 scalar hierarchy 138
multimodal 58 scenario 13
multimodal semiotic performance 44 scene 13
script 66, 72, 83
native interpreter 24 scripticals 66
Natural Language Processing (NLP) 79-80, scriptionary 72
83 second language learning 97-99
168 Subject index
second language teaching 97-99 structuralism 8, 43

semiotic action formation 43, 44 style, rhetorical 4
semiotic acts supersubject 94
— content-specific 51 symbolic signs 45-46, 87, 88, 145
— type-specific 52 Swabian vii, 1, 155
semiotic functional analysis 87
semiotic resource systems 9, 42, 43, 51 Telugu 108
sequential analysis 11, 86, 89, 94 three-part exchange 87, 99 n. 5
sequential ordering (of an interaction) 9, topic 92-94
88, 93, 95 topological-continuous variation 46, 129
Serbian 106 transcription 22, 90, 96, 99 n. 12, 100 n. 23
shifters 87 type-category 48, 51, 130
social identity 5 typification 15-16, 89, 99 n. 8
social network 18, 37, 135, 156 typological-categorial (digital) distinctions
social voice 143-144 45-46, 129
socio-cognitivism, Gumperzian 72
sociolinguistic co-occurrence rules 2, 3 Ungarinyin 58 n 4
space-time scales 142 Urdu 2, 108, 109
speech act strategy 54
speech activities 3, 4 value 43
speech communities 2, 93 variables, linguistic 2
speech events 82, 90, 97-98
stereotypes 93-94 word-object correspondence relation 45
stigmatization 93-94
Subject index 169
Author Index
Akinnaso, Niyi 17-18 41, 42, 44, 50, 53, 64, 67, 70, 71, 93, 110,
Auer, Peter 26, 67, 70, 71, 100 n. 24 112, 115, 127, 133, 135, 139, 142, 143,
Aulakh, Gurinder 99 n. 4, 100 n. 18 145, 153
Drew, Paul 27, 89, 100 n. 23
Bakhtin, Mikhail 43, 54, 55, 56, 59 n. 6, 94, Du Bois, John W. 124 n. 1
143, 144 Duranti, Alessandro 33, 63, 72
Ballim, Afzal ix, 66, 81, 118, 119, 120
Barwise, Jon 82 Eerdmans, Susan L. ix, 29 n. 1, n. 3, 114,
Bateson, Gregory 9, 27, 36, 46, 55, 67, 69, 115, 117, 118
129 Ensink, Titus 70
Bazzanella, Carla 69 Erickson, Frederick 67, 68
Berenz, Norine 72, 89, 90, 99 n. 10, 99 n.
12, 120 Fairclough, Norman 159
Berge, Kjell L. 65 Ferguson, Charles 156
Berger, Charles R. 67, 72 Field, Margaret 100 n. 27, 119
Blom, Jan-Petter 31, 66, 72, 108, 112, 156 Fillmore, Charles 13, 32
Bloomfield, Leonard 8, 106 Firth, John R. 131, 133
Blum-Kulka, Shoshana 66 Foppa, Klaus 27
Bolinger, Dwight L. 58 n. 5 Fraser, Colin 68
Bourdieu, Pierre 5, 8
Bradac, James J. 67 Gal, Susan 38 n. 1
Brown, Penelope 38 n. 1, 68, 69 Galilei, Galileo 66
Garfinkel, Harold 8, 10, 11, 19, 59 n. 6, 67,
Cacioppo, John 68 68, 72, 86, 119
Caldas-Coulthard, Carmen 159 Giles, Howard 67, 68
Cassell, Justine 72 Goddard, Cliff 83 n. 2
Castelfranchi, Cristiano 64, 68 Goffman, Erving 8, 10, 13, 14, 19, 27, 56, 59
Chomsky, Noam 136, 149 n. 6, 67, 72, 87, 114
Clark, Herbert H. 72, 99 n. 9, 120 Gomez-Imbert, Elsa 108
Coakes, Elayne 83 n. 2 Goodwin, Charles 72, 86
Cook-Gumperz, Jenny 70, 100 n. 27, 115, Goodwin, Marjorie H. 72
119, 123, 160 Greif, Irene 83 n. 2
Coulthard, Malcolm 159 Grice, H. Paul viii, 26, 32, 35, 63, 72, 86,
Couper-Kuhlen, Elizabeth 70, 73 n. 3, 100 117
n. 24 Grosjean, Michèle 64
Cowley, Stephen J. 137, 138 Grosz, Barbara 82
Grundy, Peter 72
Dascalu Jinga, Laurentia 69 Gumperz, John J. vii, viii, ix, x, xi, 1-6, 7-10,
Di Luzio, Aldo vii, viii, 14, 15, 26, 28, 31, 10-14, 15, 17-18, 19-21, 22-23, 24-25, 26-
170 Author index
27, 27-28, 29 n. 1, 29 n. 2, 29 n. 3, 31, 32, 67, 69, 70, 72, 105, 112, 114, 116, 127,
33, 35, 37, 38, 41, 42, 43, 44, 45, 46, 49, 130, 156
50, 53, 54, 55, 56, 57, 58, 59 n. 6, 63-73, Lucy, John A. 113, 121
79, 81, 82, 85-100, 107-111, 112, 115, Lyons, John 83 n. 2
116, 118, 121, 123, 127, 130, 131, 133,
135, 138, 139, 140, 143, 144, 145, 149, Martin, James R. 43
150, 151, 152, 153, 154, 155, 156 Mertz, Elizabeth 99 n. 6
Müller, Frank 70
Hall, Joan K. 100 n. 25 Murray, Stephen O. 73 n. 2
Halliday, M.A.K. 43, 47, 52, 58 n. 5, 131,
132, 138 Nöth, Winfried 49
Hanks, William F. 9, 28, 112, 114, 150
He, Agnes W. 100 n. 25 Ochs, Elinor 97
Herasimchuk, Eleanor 71 O’Connor, Mary C. 99 n. 4
Heritage, John 86 Orletti, Franca 68
Horton, William S. 83 n. 2
Hymes, Dell 3, 72, 111, 123, 149 Pallotta, Vincenzo 81
Parmentier, Richard J. 99 n. 6
Ionescu-Ruxandoiu, Liliana 15, 29 n. 2 Parret, Herman vii
Irvine, Judith 160 Peirce, Charles S. 10, 16, 49, 59 n. 6, 69, 87,
113
Jacquemet, Marco 23, 123, 152 Perry, John 82
Jacoby, Sally 97 Petty, Richard 68
Jakobson, Roman O. 87 Pike, Kenneth L. 58 n. 2, 132, 150
Jefferson, Gail 22, 95 Prevignano, Carlo L. vii, viii, x, 7, 10, 15,
16, 17, 18, 19, 21, 22, 23, 24, 25, 26, 27,
Kaal, Bertie xi 28, 41, 42, 44, 50, 53, 64, 66, 69, 71, 72, 73
Kallmeyer, Werner 150 n. 2, 93, 105, 112, 114, 115, 120, 122, 127,
Kaltman, Hannah 99 n. 4, 100 n. 18 133, 135, 139, 142, 143, 145, 146 n. 1,
Kendon, Adam 67, 68 149, 151, 152, 153, 154, 157-158, 159
Kerbrat-Orecchioni Catherine 70
Keysar, Boaz 83 n. 2 Raskin, Victor 72
Kiesler, Sara 83 n. 2 Rawlins, William K. 69
Koerner, E.K.F. 66 Roberts, Celia 97, 98, 100 n. 18
Rumsey, Alan 45, 58 n. 4
Labov, William 37
Lakoff, Robin 32 Sacks, Harvey 32, 67, 72, 95
Langacker, Ronald W. 48, 51 Sapir, Edward 8, 43
Lave, Jean 120 Sarangi, Srikant 72, 98
Leach, Edmund 155 Saussure, Ferdinand de 8, 10, 43, 158
Leech, Geoffrey viii, 63 Schank, Roger C. 83
Lemke, Jay 41, 43, 44, 49, 51, 52, 53, 54, 55, Schegloff, Emanuel A. 72, 95
129, 133, 134, 136, 138, 139 Schiffrin, Deborah 23, 68, 72, 85, 86, 98
Levinson, Stephen C. viii, 31, 52, 55, 64, 65, Schutz, Alfred 15, 99 n. 8
Author index 171
Searle, John R. 32 Tyler, Andrea 100 n. 19

Selting, Margret 70, 96, 124 n. 1
Silverstein, Michael 36, 45, 47, 56, 59 n. 6, Uhmann, Susanne 100 n. 24
87-88, 99 n. 6, 99 n. 12, 113, 120, 150, Urban, Greg 120
151, 157, 158
Sinclair, John 21 Van Leeuwen, Theo 44, 51, 52, 128
Sproull, Lee 83 n. 2 Vanelli, Laura 69
Stati, Sorin 28 n. 1, 68 Vico, Giambattista 6
Svennevig, Jan 67 Voloshinov, Valentin N. 123
Szymanski, Margaret 100 n. 27, 119
Weigand, Edda 17
Tannen, Deborah 38 n. 1 Weizman, Elda 66
Thibault, Paul J. viii, x, 41, 43, 44, 49, 50, Wenger, Etienne 120
51, 52, 53, 54, 55, 69, 73 n. 2, 120, 121, Whorf, Benjamin L. 43, 152
128, 134, 139, 143, 144, 145, 150, 152- Wilks, Yorick 66, 81, 82
153, 159 Wittgenstein, Ludwig 58 n. 1
Thomas, Jenny 72 Wodak, Ruth 159
Toribio, Almeida 87
Tsui, Amy 87, 99 n. 5 Young, Linda 23
Turing, Alan M. 72 Young, Richard 100 n. 25

Language and Interaction

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Language and Interaction

Hochgeladen von

Copyright:

Verfügbare Formate

<DOCINFO

TITLE "Language and Interaction: Discussions with John J. Gumperz"

SIZE HEIGHT "220"

Language and Interaction

John J. Gumperz, Professor Emeritus of Anthropology, Uni-

Photograph by Paul Bishop with kind permission.

John Benjamins Publishing Company

of American National Standard for Information Sciences – Permanence

Library of Congress Cataloging-in-Publication Data

Language and interaction : discussions with John J. Gumperz / edited by Susan

P85.G84 L36 2002

© 2003 – John Benjamins B.V.

Bio-bibliographical note 163

The aim of the present collection of papers is to highlight some aspects of

The present is a revised edition of a volume which was issued as a pre-

Bertie Kaal of John Benjamins provided us with constant editorial advice

Presenting John J. Gumperz

As a linguist and anthropologist, John Gumperz has successfully combined

According to this theory, the interpretation of meaning is interactively negoti-

ings, law proceedings, formal interactions in the work place or at school. In

attention to the fact that discrimination in interaction has a linguistic dimen-

A discussion with John J. Gumperz1

Carlo L. Prevignano and Aldo di Luzio

in the ethnography of communication. The key insight here is that ethno-

course Strategies, I began to realize that Saussurean phonological and gram-

linguistic, social and cultural/ideological forces and governed or constrained

cal signs in interpretation, it becomes necessary to distinguish between mean-

following lines. Analysis begins with turn-by-turn scanning at two levels of

presuppositions about mutual rights and obligations, as well as on ideologies

C.L.P.: As to the metapragmatic activity, I think it is important in order to

I replied, “Sure, come to the oYce,” whereupon he went on to say in what

other fellows of King’s, not to the public at large. Understanding, as struc-

C.L.P.: Or corpus linguistics, in this case for pragmatic analysis.

graphically-based analysis and information on the relationship between lin-

C.L.P.: I think that as an analyst, an imperfect analyst, I always have to resist

valid ethnographic information can lead to more sophisticated and informa-

C.L.P.: So, in your sense, pragmatic creativity can be a sort of violation of

Eerdmans, Susan L. “A review of John J. Gumperz’s current contributions to Interactional

Contextualizing “contextualization cues”

I would like to focus my remarks on John Gumperz’s notion of a contextua-

not to be bridged completely with the apparatus at hand — implicatures,

1. Context-importation by conventional coding devices

Examples would be presupposition triggers like deWnite articles, expressions

2. Context-invocation by inference alone

Classical Gricean implicatures would be the prototype here:

where in order to infer a (negative) response to the invitation, it is necessary to

3. Context-invocation by “cue” or “Xag”

Table 1. The opposition between central and peripheral information.

Form lexico-syntactic particles, modiWers,

This set of alignments is the intuition in the Batesonian analysis of communi-

channel (a constellation of e.g. kinesic, prosodic and lexical features).

pessimistic story about how elites guard access to social advancement.

My perspective is in broad sympathy with Gumperz’s attention to “the de-

2. Social meaning-making: Resource systems and practices

words, we need a complementary account of the social semiotic resource

speciWable level of delicacy, unique to the particular context.

3. Context-dependent and context-independent meanings

Gumperz’s solution to these issues lies in what he calls a “move from a

not to be so readily segmentable into discrete particles or constituents of

4. Discourse, context, and meaning-making practices: Towards an

Gumperz draws attention to the shifting nature of participant interpretation

The three categories are analytically, though not constitutively, separable

I. Indexical meaning-making practices

Indexicality, or “grounding” (Langacker 1987:126–8) refers to the fact that all