Stancetaking in Discourse Subjectivity Evaluation Interaction PDF

Stancetaking in Discourse
Pragmatics & Beyond New Series (P&BNS)

Pragmatics & Beyond New Series is a continuation of Pragmatics & Beyond and
its Companion Series. The New Series offers a selection of high quality work
covering the full richness of Pragmatics as an interdisciplinary field, within
language sciences.
Editor
Andreas H. Jucker
University of Zurich, English Department
Plattenstrasse 47, CH-8032 Zurich, Switzerland
e-mail: [email protected]
Associate Editors
Jacob L. Mey Herman Parret Jef Verschueren
University of Southern Belgian National Science Belgian National Science
Denmark Foundation, Universities of Foundation,
Louvain and Antwerp University of Antwerp
Editorial Board
Shoshana Blum-Kulka Susan C. Herring Emanuel A. Schegloff
Hebrew University of Indiana University University of California at Los
Jerusalem Angeles
Masako K. Hiraga
Jean Caron St.Paul’s (Rikkyo) University Deborah Schiffrin
Université de Poitiers Georgetown University
David Holdcroft
Robyn Carston University of Leeds Paul Osamu Takahara
University College London Kobe City University of
Sachiko Ide
Foreign Studies
Bruce Fraser Japan Women’s University
Boston University
Catherine Kerbrat- Sandra A. Thompson
University of California at
Thorstein Fretheim Orecchioni
Santa Barbara
University of Trondheim University of Lyon 2
John C. Heritage Claudia de Lemos Teun A. van Dijk
Pompeu Fabra, Barcelona
University of California at Los University of Campinas, Brazil
Angeles
Marina Sbisà Richard J. Watts
University of Berne
University of Trieste
Volume 164
Stancetaking in Discourse. Subjectivity, evaluation, interaction
Edited by Robert Englebretson
Stancetaking in Discourse
Subjectivity, evaluation, interaction
Edited by
Robert Englebretson
Rice University
John Benjamins Publishing Company

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

Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data
Stancetaking in discourse : subjectivity, evaluation, interaction / edited by Robert

Englebretson.
p. cm. -- (Pragmatics & beyond new series, issn 0922-842X ; v. 164)
Includes bibliographical references and index.
1. Discourse analysis. 2. Conversation analysis. I. Englebretson, Robert.
P302.S65 2007
401'.41--dc22 2007026354
isbn 978 90 272 5408 5 (Hb; alk. paper)
© 2007 – 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
Table of contents
Acknowledgements vii
Stancetaking in discourse: An introduction 1

Robert Englebretson
Using a corpus to investigate stance quantitatively and qualitatively 27

Susan Hunston
Linking identity and dialect through stancetaking 49

Barbara Johnstone
Grammatical resources for social purposes: Some aspects

of stancetaking in colloquial Indonesian conversation 69
Robert Englebretson
Subjective and intersubjective uses of generalizations

in English conversations 111
Joanne Scheibman
The stance triangle 139

John W. Du Bois
The role of I guess in conversational stancetaking 183

Elise Kärkkäinen
Stance markers in spoken Finnish: Minun mielestä

and minusta in assessments 221
Mirka Rauniomaa
Stancetaking as an interactional activity: Challenging the prior speaker 253

Tiina Keisanen
Positioning and alignment as activities of stancetaking in news interviews 283

Pentti Haddington
Name index 319

Subject index 321
Acknowledgements
The papers in this volume have their genesis in work presented at the 10th Bien-
nial Rice Linguistics Symposium, held March 31 through April 3, 2004, at Rice
University. I would like to thank each of the presenters and attendees for mak-
ing this into a rich and fruitful conference, and especially for the lively and en-
gaging roundtable discussion that concluded the conference. Presenters (listed
in alphabetical order) were John W. Du Bois, Robert Englebretson, Pentti Had-
dington, Susan Hunston, Barbara Johnstone, Elise Kärkkäinen, Tiina Keisanen,
Mirka Rauniomaa, Geoffrey Raymond, Joanne Scheibman, and Robin Shoaps.
We gratefully acknowledge the following organizations for funding the sympo-
sium: Rice University Department of Linguistics; the Stance Research Group at
the University of Oulu, Finland; and the Rice University Humanities Research
Center (formerly the Center for the Study of Cultures). I would also like to thank
the graduate students and faculty in the Rice Linguistics Department for provid-
ing logistical support and a warm welcome (Houston weather not withstanding!)
to our guests.
We would also like to express our gratitude to the two anonymous reviewers
for their input and comments, which have improved the overall quality of the
manuscript.
For financial support in the completion of this project, I gratefully acknowl-
edge the Rice University Office of the President (Presidential Research Award
2006–2007).
I would especially like to thank both Mara Henderson and Michelle Morri-
son for their painstaking and diligent editorial assistance at various stages of the
manuscript.
Stancetaking in discourse
An introduction
Robert Englebretson
Rice University
1. Introduction
During the initial few years of the twenty-first century, scholarship in linguis-
tics and related disciplines has witnessed a notable upsurge of interest in stance.
Several published monographs explicitly reference stance in their titles (Gardner
2001; Hunston and Thompson 2000; Kärkkäinen 2003; Mushin 2001; Wu 2004),
special issues of journals have focused on this topic (Berman 2005; Macken-
Horarik and Martin 2003), panels and symposia have been devoted to various
perspectives on stance (Englebretson 2004; Jaffe 2004; Shoaps and Kockelman
2002), and numerous journal articles have dealt with this topic either directly or
indirectly. This broad array of research represents a convergence; an intersection
of subdisciplines within linguistics (among e.g., corpus linguistics, systemic-func-
tional linguistics, discourse-functional linguistics, cognitive linguistics, sociocul-
tural linguistics, and interactional linguistics), and it highlights a set of overlap-
ping interests with closely-allied fields such as anthropology, social psychology,
education, and sociology. As this heterogeneous range of research implies, and as
I will argue throughout this chapter, stance is by no means a monolithic concept.
Definitions and conceptions of stance are as broad and varied as the individual
backgrounds and interests of the researchers themselves. But what is notewor-
thy about the focus on stance from all of these different perspectives, is that it
marks an orientation toward conceiving of language in terms of the functions
for which it is used, based on the contexts within which it occurs. Research on
stance, however this term is defined, represents an ongoing trend toward under-
standing the full social and pragmatic nature of language, as it is used by actual
speakers or writers to act and interact in the real world. The present volume joins
this burgeoning field, offering new insights into the sociocultural, interactional,
Robert Englebretson
and performative nature of language by addressing stancetaking from a range of

perspectives in natural discourse.
A critical observer of the recent preponderance of work on stance may won-
der what, exactly, the term stance refers to in the first place, and to what extent
these various approaches may converge with each other overall.1 There are at least
two ways in which the use of this term can be problematic. First, while individual
researchers do tend to operationalize stance within their own work, definitions
and understandings are not necessarily shared as common ground from one
scholar to the next. Thus two authors may use stance to encompass two seem-
ingly disparate types of phenomena. Secondly, the converse situation is also true;
researchers who appear to be working on similar issues may in fact use different
terminology to cover what appears to be the same types of phenomena. For ex-
ample, one scholar may use stance to encompass what other scholars refer to as
subjectivity, while other language researchers may prefer the term evaluation and
may wish to eschew the label stance altogether.
The lack of a consistent definition of stance poses a conundrum for anyone
wishing to summarize the state of the field in an introduction such as the present
chapter. Such a writer is faced with essentially two possibilities. One approach
would be to present a clearly operationalized definition of stance, and consequent-
ly to exclude from the purview of stance research anything which does not meet
this definition. Such an approach would border on academic imperialism, and
would likely end up excluding much of what has been published under the rubric
of stance research, including many of the papers in the present volume. Prescrib-
ing a narrow definition would needlessly fragment and limit this interdisciplinary
field of research at a time when it has only just begun to emerge and take shape.
A second, more inclusive approach would be to adopt a broader, ethnographi-
cally-informed conception of stance; to seek to understand how stance research-
ers conceive of stance in their own work, and to begin to develop a model which
recognizes the heterogeneous and multifaceted nature of stancetaking. This inclu-
sive approach has the advantage of promoting dialogue among stance researchers
and encouraging interplay among ideas, and this is the approach I have chosen
to adopt here, both in this introductory chapter and in the volume as a whole. A
reader of this volume who wishes a clear-cut, operationalized definition of stance
will no doubt find it challenging to follow the disparate threads of ideas as they
coalesce throughout the various approaches represented here. However, I believe
this is the most realistic and fruitful way to approach the current, burgeoning
state of stance research. The overall purpose of this volume is to explore how
it is that speakers (and writers) actively engage in taking stances in natural dis-
course. The understanding of stance that emerges is indeed heterogeneous and
variegated, but always intimately bound up with the pragmatic and social aspects
Introduction
of human conduct, while highlighting the situated, pragmatic, and interactional

character of stancetaking. We offer the present volume as an exploratory work in
stance research, and we hope the ideas and claims in the papers will spark further
inquiry and new directions in investigating and refining this field.
The remainder of this introductory chapter is organized as follows. Section
2 steps back from academically-oriented approaches to stance, to discuss the
semantics and pragmatics of the term stance from a metalinguistic perspective.
Based on two corpora of present-day English, I outline trends in the meaning of
stance, as observed from how people actually use it in their speaking and writ-
ing. I suggest that the diverse nature of how stance is conceptualized in everyday
language use impacts and informs how language researchers have conceptual-
ized and appropriated it in academic contexts as well. Section 3 returns to the
question of how stance has been approached within language research. I survey
recent work relevant to stance, and thereby seek to contextualize the present vol-
ume within the current intellectual climate of stance-related scholarship. Section
4 briefly introduces each of the contributions to this volume.
2. Stance in discourse
The title of the present volume, Stancetaking in Discourse, entails several proposi-
tions. It presupposes first that there is a conceptual entity known as stance, which
we can observe, investigate, research, and write about. Secondly, it suggests that
stances are something that people actively engage in (i.e., stancetaking is a gerund,
based on an object-incorporation of stance and the active verb take). Finally, it
claims that stancetaking happens in discourse – in language in its natural habi-
tat – and is thus best studied within this context. The present section takes up the
first proposition just mentioned – the question of the conceptual entity known as
stance – by investigating the meaning of the word stance in order to more fully
come to terms with stance as an object of research. In order to come to an un-
derstanding of what stance may be understood to be, this section approaches the
topic from an ethnographic perspective in everyday language. In other words, set-
ting aside for now the case of academic language-related stance research to which
we shall return in Section 3, when speakers and writers use the term stance in
their naturally-occurring speaking and writing, what do they mean? This section
seeks an answer to this question by presenting a brief quantitative and qualitative
overview of the term stance as observed in two corpora of present-day English.
This approach takes seriously Hunston’s (this volume) call for both quantitative
and qualitative methodologies in the investigation of stance, and it also serves to
prefigure the kinds of analyses undertaken throughout the rest of this volume.
Robert Englebretson
A metalinguistic analysis of stance in everyday language is crucial in com-

ing to terms with the kinds of activities implicated in stancetaking. Since there is
no agreed-upon definition of stance from an academic perspective, the question
naturally arises as to why it is that stance researchers have adopted this term to
refer to such a broad and variegated set of phenomena. I would suggest that these
conceptions of stance are based directly on the wide range of meanings and activi-
ties associated with this lexeme in everyday language use. In other words, there
is a reflexive relationship between stance as a concept in language research, and
stance as a word in English more generally; the everyday conceptions of stance
serve as a window into how this term has been appropriated by linguistic, social,
and interactional research. The rather obvious fact that stance researchers are also
language users suggests that questions of how often, in what situations, and in
what kinds of interactional and collocational contexts stance is used are all highly
relevant in understanding how stance is conceptualized more specifically in re-
search contexts as well. This is firmly in keeping with a usage-based perspective
on language (cf. Barlow and Kemmer 2000; Bybee 2006; inter alia), which takes as
its starting point the idea that language use shapes language form and meaning.
As I will demonstrate in the present section, the heterogeneity of stance as an ob-
ject of research, and the varying uses of this term by scholars, are clearly signified
in the quantitative and qualitative analyses of stance in language use.
The data for this section come from two corpora: the four volumes of the San-
ta Barbara Corpus of Spoken American English (SBCSAE) (Du Bois et al. 2000;
Du Bois et al. 2003; Du Bois and Englebretson 2004, 2005) and the British Na-
tional Corpus, World Edition (BNC) (BNC 2001).2 All SBCSAE examples in this
section are excerpted from the corpus directly. For the BNC data presented in this
section, all concordance counts and collocations were obtained by using VIEW
(Variation in English Words and Phrases), the BNC online search tool provided
by Mark Davies at Brigham Young University: http://view.byu.edu.3 In addition
to representing the English of the United States versus the English of the United
Kingdom, these two corpora are vastly different from each other in terms of size,
scope of genre, and transcription detail. The SBCSAE contains approximately
249,000 words at present;4 the BNC is over 400 times larger, comprising approxi-
mately 100 million words.5 The SBCSAE focuses entirely on spoken language,
primarily (but not exclusively) on face-to-face conversation; the BNC consists of
a written component (90% of the corpus) and a spoken component (10% of the
corpus). The SBCSAE is transcribed using the intonation unit-based discourse
transcription system proposed by Du Bois et al. (1993), and contains time-aligned
audio for each transcript; the spoken component of the BNC uses an undifferenti-
ated orthographic transcription system with minimal attention to interactional
or intonational details, and the audio files are neither included nor time-aligned
Introduction
with the transcripts.6 Finally, while the SBCSAE is untagged, the BNC is tagged
for parts-of-speech and sentences, and can thus be used with parsers and other
automated tools. Obviously, each corpus is more suitable for some types of re-
search purposes than it is for others. Because of its close focus on interactional
detail and the availability of the audio files, the SBCSAE is especially useful for in-
teractional, ethnographic, acoustic-phonetic, and conversation-analytic research
(and for traditional corpus work too, of course, if the research does not require
data from written English, and if large-scale frequency counts are not a focus of
concern). Due to its massive size, tags, and genre-based orientation, the BNC is
an especially useful resource for large-scale quantitative computational and text
work (as long as the researcher is not seeking interactional detail, prosodic infor-
mation, or audio files in the spoken component). Given the somewhat comple-
mentary natures of these corpora, the two together provide an excellent vehicle
for qualitative and quantitative analyses of stance. I will begin this discussion in
Section 2.1 by presenting a broad quantitative overview of the rate of occurrence
of stance in both corpora, and in sub-genres within the BNC. Then, in Section
2.2, I will move to a qualitative investigation of the occurrences of stance in the
SBCSAE. Section 2.3 discusses the adjectives in the BNC that most frequently
collocate with stance, in order to analyze the semantic prosodies (cf. Hunston this
volume) that are most typically associated with this term. Section 2.4 summarizes
the findings of this investigation and outlines what it has to contribute to our un-
derstanding of the term stance specifically, as well as to our general conception of
stance in language research, which is the topic of Section 3.
2.1 Quantitative findings
This section addresses the quantitative distribution of stance in both corpora. We

observe that tokens of stance are quite rare in both speech and writing, but occur
more frequently in writing than in speech. We also find that stance shows skewed
distribution across registers of the BNC. Table 1 below, presents the overall occur-
rence of tokens of the lemma stance (i.e., the singular form stance and the plural
form stances combined) in both corpora.
Table 1 shows two important quantitative findings about stance. First, it is a
fairly infrequent word in both corpora. It occurs only three times in the SBCSAE,
giving a rate of occurrence of 1 token per every 83,000 words of talk (and making
further quantitative analysis of this lexeme impossible in this corpus).7 The lemma
stance occurs a total of 1,835 times in the combined written and spoken compo-
nents of the BNC, with a rate of occurrence of roughly 1 token per every 53,198
words of text – also relatively rare, although more frequent than in the SBCSAE.
Robert Englebretson
Table 1. Tokens of stance in the SBCSAE and BNC

Corpus Tokens Total words Rate of occurrence
SBCSAE 3 249,000 1 per 83,000 words
BNC (Spoken Component) 42 10,334,947 1 per 246,070 words
BNC (Written Component) 1,793 87,284,364 1 per 48,681 words
BNC (Total) 1,835 97,619,311 1 per 53,198 words
Secondly, this table shows that stance occurs far more frequently in writing than it
does in speech, at a greater-than five to one ratio. When the BNC is broken down
by written versus spoken texts, the rate of occurrence of stance in the written
component is 1 token per every 48,681 words of text, as contrasted with roughly
1 token per every 246,070 words in the spoken component.8 Thus stance occurs
over five times more frequently in writing than it does in speaking.
Furthermore, when stance does occur in the BNC, it is broadly correlated
with formal registers. The three text types with the highest rate of occurrence
of stance in the BNC are university essays, newspaper writing on science, and
non-academic writings on law, politics, and education. In the spoken component
of the BNC, the two text-types with the highest rate of occurrence of stance are
scripted speeches and social science lectures. Interestingly, there is not a single
token of stance in the entire 4-million word BNC sample of conversation. I shall
return to the question of what exactly these distributional factors tell us about
stance in Section 3.4 below, but it is necessary to first address this term from a
qualitative and interactional perspective.
2.2 Qualitative analyses
As shown in Table 1 in the previous section, the SBCSAE contains only three to-
kens of stance. The present section provides a qualitative analysis of each of these
tokens within their interactional contexts-of-use. These analyses bring to light five
key conceptual principles of stance as follows: (1) stancetaking occurs on three
(often overlapping) levels – stance is physical action, stance is personal attitude/be-
lief/evaluation, and stance is social morality; (2) stance is public, and is perceivable,
interpretable, and available for inspection by others (cf. Du Bois this volume); (3)
stance is interactional in nature – it is collaboratively constructed among partici-
pants, and with respect to other stances (Scheibman, this volume, makes a similar
point about the “relational” nature of stance in discourse); (4) stance is indexical
(cf. Haviland 1989; Silverstein 1976), evoking aspects of the broader sociocultural
framework or physical contexts in which it occurs; (5) stance is consequential –
i.e., taking a stance leads to real consequences for the persons or institutions in-
Introduction
volved (cf. Du Bois, this volume, for the relationship between stancetaking and
responsibility). In the remainder of this section, I will demonstrate how each of the
three SBCSAE excerpts illustrates these five principles. In Section 2.3, I will present
quantitative collocational evidence from the BNC which support these qualitative
findings. Overall, I suggest that a qualitative understanding of what stance means
in talk-in-interaction serves as a window into the academic concept of stance –
both in the papers in this volume, and in language-related research on stance more
generally. As we examine stance from a qualitative perspective in everyday talk, the
reflexive relationship between these uses and the conceptualization of this term in
academic stance research becomes apparent. We now turn to an analysis of each of
the three stance tokens in the SBCSAE.
Example (1) comes from a recording of a judo class. In this speech event
(SBC057, entitled Throw Me) the sensei, Nick, is teaching and demonstrating
Hane-Makikomi (springing wraparound throw). This excerpt begins as Nick is
explaining the kinds of situations in which Hane-Makikomi would be useful.9
(1) (Throw Me SBC057: 948.362-966.593)

1 NICK: ... But I like the Hane-Makikomi,
2 ... (H) when it’s used on a player (Hx),
3 ... who’s fighting you in that,
4 → ... extre=me (Hx) defensive stance.
5 ... (H) While he’s down in a deep jigo.
6 .. Preferably,
7 ... with his left foot forward.
8 ... Pressing against you with that left hand.
9 ... Keeping you out.
The term stance appears at the arrow in line 4 of this example, extreme defen-
sive stance. This excerpt presents a hypothetical example – a description of when
Hane-Makikomi would be a desirable throw. The interactants are you (the judo
practitioner who will be using Hane-Makikomi) and he (the opponent who is
fighting you in that extreme defensive stance).
The textual and physical contexts of stance in this excerpt illustrate each of the
five principles outlined above. First, in this case, the use of stance here is clearly a
physical act. As described in Nick’s subsequent explanation, this extreme defen-
sive stance refers literally to a physical way in which the opponent is positioning
his body, down in a deep jigo (line 5) (jigo is a judo term referring to ‘self-defen-
sive posture’), with his left foot forward (line 7), Pressing against you with that left
hand (line 8). All of these descriptions refer to the opponent’s stance, i.e., body
posture. Secondly, the physical stance of this hypothetical opponent is public,
Robert Englebretson
and is indeed being inspected and interpreted. In this case, Nick observes and
inspects the hypothetical opponent’s stance and describes it in detail. This kind
of overt inspection of stance is, after all, part of assessing one’s opponent during
a judo match. He then interprets the stance as Keeping you out (line 9). Thirdly,
this excerpt represents the interactionality inherent in taking a stance. On one
level, the physical details of this particular stance require two individuals: you
and he (your opponent). You must be there in order for the opponent to be press-
ing against you with that left hand (line 8). Similarly, it is your previous offensive
moves that have led to the opponent’s current extreme defensive stance, and it is
the opponent’s extreme defensive stance that will lead to your subsequent use of
Hane-Makikomi. Fourthly, stance is indexical. Here, the physical stance points to
(indexes) something beyond the textual and physical context: namely, it indexes
the specific knowledge systems of judo, as well as the sociocultural background in
which this martial art is embedded. This stance of a deep jigo has specific mean-
ing within the practice of judo; i.e., seeing the opponent in this physical stance
calls up a host of understandings about the opponents intentions, the opponent’s
probable next actions, and your best move. Finally, stance has consequences. In
this case, the opponent’s stance has the direct and physical consequence of elicit-
ing Hane-Makikomi. And this relationship between stance and consequence is
indeed the upshot of Nick’s instruction in this excerpt. To roughly paraphrase:
if your opponent takes a physical stance like this, then you respond with Hane-
Makikomi.
While example (1) illustrates physical stance, the following two examples
illustrate personal and moral stance. These remaining two tokens of stance in
the SBCSAE both come from the same speech event (SBC035 entitled Hold my
Breath). The relevant background information for understanding these excerpts is
as follows. Stephanie (Steph) is a senior in high school and is presently applying to
colleges. Gail is Stephanie’s older cousin, who is currently a college student. Patty
is Stephanie’s mother, and Maureen (Maur) is Gail’s mother.10 The conversation
is taking place at a family get-together. In example (2), Gail is describing an event
from the past, her application interview to attend a private Catholic college in the
northeast – a college to which she was not admitted.
(2) (Hold my Breath SBC035: 285.730-303.681)

1 GAIL: .. You know what,
2 they asked me some really weird [questions] though.
3 MAUR: [I know].
4 I know[2=2].
5 PATTY: [2They2] [3did3],
6 GAIL: [3Like3] they --
Introduction
7 I expected to go in there= and talk about,

8 you know my li=fe,
9 and what I wanted out of a schoo=l,
10 and what I expected from [the=m and,
11 PATTY: [So what did they ask you].
12 → GAIL: (H)] They asked me .. what my stance was abou- on abortion?
13 .. [2(H)2]
14 PATTY: [2Boy that2] was a dumb question,
15 [3that’s one --
16 GAIL: [3Boom.
17 I was like .. (TSCH)3],
18 PATTY: That’s a contro3]versial question,
In this excerpt, the stance token occurs in line 12 at the arrow, They asked me what
my stance was on abortion. Unlike in example (1), this stance does not concern
physical body posture, but rather the speaker’s beliefs about, attitudes toward,
and evaluation of a controversial (line 18) and very personal (as cited later in the
transcript – not shown here) moral issue. Because of the similarities between this
and the next example, I will discuss them both together after presenting example
(3), which picks up 25 seconds later in the speech event from the end of example
(2). Prompted by Gail’s story of not being admitted to a college after an interview,
Stephanie recounts the following experience of her friend Lee who was waitlisted
under similar circumstances at a different college.
(3) (Hold my Breath SBC035: 328.955-365.521)

1 STEPH: (H) Lee applied to Williams.
2 And she got waitlisted.
3 .. (H) And Williams used to be real=ly lib- --
4 ... (H) liberal whatever.
5 .. But lately,
6 .. the= information they’re sending out is,
7 (H) <VOX a ni=ce value= community uh,
8 → (H) moral stance,
9 da-da-da-da-da [I mean VOX>,
10 PATTY: [There’s nothing wrong with that,
11 STEPH: (H) Well no,
12 but what they’re do- --
13 PATTY: that’s important].
14 STEPH: But],
15 but m- Mom,
10 Robert Englebretson
16 what I’m [saying is],

17 PATTY: [I’m listening].
18 STEPH: (H) ... They had,
19 ... years ago,
20 .. they had <VOX the sit-ins VOX>,
21 .. everything you know,
22 (H) real liberal,
23 .. now they’re going to conservative,
24 and when she was interviewed for it,
25 (H) they= .. took her answers .. very .. slowly,
26 and they waitlisted her,
27 (H) she had the grades,
28 .. but she didn’t have the political views.
The stance token in this excerpt occurs in line 8 at the arrow, moral stance. As
in the previous example, this use of stance refers to beliefs, attitudes, and values.
However, unlike in the previous example, the stance here in example (3) is attrib-
uted to a general institution (Williams), rather than to a specific person (Gail) as
in example (2).
Taken together, examples (2)–(3) illustrate the five principles of stancetaking
described earlier. First, stance in these examples is depicted as a personal belief or
attitude, and likewise as a social value – e.g., moral stance in line 8 of example (3).
Secondly, the observable, interpretable, and public nature of stance is highlighted
in both cases. In example (2), Gail’s stance was specifically elicited in the con-
text of an interview. By the very nature of an interview, her stance is presumably
being scrutinized, evaluated, and interpreted by the interviewer(s), for purposes
of making a decision on her admission to the college. Similarly, in example (3),
Stephanie is observing and criticizing the conservative moral stance of the col-
lege, at the same time that her mother is defending it. Stephanie’s criticism of the
college’s stance is suggested by the mocking-sounding prosody in lines 7–9, and
by the overall negative tone of her claim that the college has shifted from liberal
to conservative. But the most important evidence that Stephanie is criticizing the
college’s stance is displayed in the uptake by her mother. Patty likewise interprets
Stephanie’s comments as criticism, and in turn seeks to defend the college’s stance,
There’s nothing wrong with that (line 10). Subsequently, Patty herself evaluates the
college’s stance, that’s important (line 13). The criticism, defense, and evaluation
displayed in this sequence illustrates that stance is indeed observable and avail-
able for interpretation. Thirdly, both excerpts illustrate the interactional nature of
stance. In example (2), the upshot of Gail’s story is the juxtaposition of her stance
on abortion with the (implied) stance of the private Catholic college. In example
Introduction 11
(3), Stephanie’s point is that the institution’s stance was apparently at odds with
the stance of her friend Lee. In both cases, it is the dialectic of conflict between
the stances of the individual and the stances of the institution that lead to the
overall realization of the stance more generally – just as in the case of the physical
stance discussed in example (1), where the extreme defensive stance is described
in terms of the positioning of two bodies with respect to one another. Fourthly,
both excerpts demonstrate the indexical nature of stance. In example (2), Gail’s
stance on abortion, which is never overtly stated here, is seen as indexing wider
social morals and values; a so-called “anti-abortion” or “pro-life” stance would
stereotypically index Gail as likely being a person who supports more conserva-
tive political views and agendas, while a so-called “pro-abortion” or “pro-choice”
stance would stereotypically index and imply that she holds a wider array of more
liberal political views. In example (3), the college’s “moral stance” likewise indexes
a range of other stances, a nice value community (line 7) and an unspecified list
of other conservative views insinuated by Stephanie’s da-da-da-da-da (line 9). Fi-
nally, in both excerpts, stance is consequential. The implications of example (2)
are that Gail’s stance on abortion may have led her to not gain admission to this
particular school, while in example (3), Stephanie suggests that the lack of align-
ment between Lee’s views with the stance of the college may have led to Lee being
waitlisted, she had the grades, but she didn’t have the political views (lines 27–28).
The qualitative analyses of these three excerpts from the SBCSAE have illus-
trated five general principles of stancetaking in discourse. In the following sec-
tion, we will return to the BNC and examine how collocational evidence supports
these qualitative findings as well.
2.3 Collocational evidence
The previous section has illustrated five general themes observed in stancetaking
in discourse: stance is physical/personal/moral, stance is public and interpretable,
stance is interactional, stance is indexical, and stance is consequential. This sec-
tion provides a brief overview of the adjectives in the BNC that are immediate
collocates with (i.e., occur one word to the left of) stance. An analysis of the kinds
of adjectives that typically occur with stance leads to a richer understanding of the
meaning of this term, and supplements the quantitative and qualitative findings
presented in the previous two sections.
The following table shows the 20 adjectives in the BNC that collocate with
stance more than five times in the corpus.
Table 2. Adjectives in the BNC collocating with stance (frequency > 5)
Adjective Number of tokens
political 37
aggressive 20
moral 16
upright 15
tough 13
critical 12
neutral 10
positive 10
forward 10
public 10
negative 9
basic 9
particular 8
left 8
ideological 7
conservative 7
anti-abortion 6
previous 6
different 6
right 6
This table shows that there are only 20 adjectives in the BNC that serve as imme-
diate collocates of stance with a frequency greater-than five tokens apiece. These
adjectives are illustrative of how stance is conceptualized in naturally-occurring
speech and writing, and I will now turn to a summary of these findings with re-
spect to the qualitative observations from the previous section.
First, collocates of stance reflect the physical, evaluative, personal, and moral
dimensions of stance, and these categories often overlap. A close inspection of the
larger co-text shows that there are four adjectives in this list that refer to physical
stance only: upright, forward, left, and right (with the exception of one token of
right meaning ‘morally correct’). For example: “Fair-haired, tall and with a dis-
tinctive upright stance, Geoff ’s powerful running, strength, and cultured style…”
(B2H); “For the evolving herbivores the advantage of an upright stance was soon
complemented by the ability to rear up…” (C9A); “From this position the begin-
ner steps into the forward stance, in which about 60 per cent of the body weight is
on…” (GVF); “You take up a left stance and line yourself up so that your left foot
is in front of the opponent…” (A0M); “The attacker steps forward into the right
stance to deliver a right lunge punch.” (GVF). One token of right is evaluative in
nature and reflects moral stance: “…exotic religious sect. He was both eager to
Introduction 13
adopt the right stance and unnerved by the strangeness of it.” (FAJ). Other evalu-
ative stance collocates include critical, positive, neutral, and negative. For example:
“…loss of support for the Greens was attributed to their critical stance on unifica-
tion…” (HL2); “Clearly, this kind of legislation reflects a more positive stance to-
wards older workers and their role in the labour market.” (B01); “I have to report a
very negative stance by the British Mountaineering Council to our association…”
(ECG). Adjectives that highlight the personal and moral nature of stance include:
moral, ideological, conservative, and anti-abortion. For example: “But this does not
mean that it takes the conservative stance of necessarily accepting existing defini-
tions of crime.” (CRX); “…the anti-abortion lobby had apparently been indicat-
ing the ‘weak’ moral stance of protestants on the issue.” (A07); “…at a time when
Bush adopted an anti-abortion stance. And many American women see Hillary
Clinton as much more…” (CEK). Two collocates, tough and aggressive, suggest
that stance may often imply a high degree of severity or strength, and this applies
to both physical and personal/moral stance. For example, physical stance: “There-
fore, Piaroa boys are not constrained into learning the aggressive stance of young
warriors to fight men of this world…” (CJ1); personal/moral: “…men are useful
but disposable. The victims of their aggressive stance are their children.” (CH1);
and “it was necessary for the Soviet Union to adopt a more aggressive stance if it
were to maintain credibility as a rival to Washington…” (G1R).
The second principle observed in the previous section, namely that stance is
public, is overtly supported by the collocational evidence as well – in particular by
the adjectives political (the most frequent adjective collocate of stance in the BNC)
and the adjective public itself. Consider for example: “…the group have attracted
most attention for their confrontational political stance, most pertinently summed
up on their recent mail order-only single and the LP’s strongest track, ‘Fuck The
Right To Vote’.” (CGB). Here, the music group’s political stance is apparently suffi-
ciently well-known as to attract attention, and is summed up (publicly) on a single
from a record album. Similarly, in “A new Archbishop of Canterbury was selected
who took a public stance against clergy in gay and lesbian relationships…” (C9S),
the public nature of the stance is highlighted by the very adjective itself.
The third principle, namely the interactional nature of stance, is difficult to
assess in terms of the collocational evidence alone. However, it does receive sup-
port from the collocates basic, particular, previous, and different. Each of these
adjectives implies some sort of comparison with other stances: basic as opposed
to more complicated, particular as opposed to general, previous as opposed to
current and future, and different, which, by virtue of its very meaning, sets two
stances in opposition to one another. Consider the use of different stance in the
following BNC excerpt: “Geraldine Pederson-Krag, in a paper written in 1951,
took a different stance from her psychoanalytic contemporaries when she ana-
lysed the system of mass production.” (CBH). In this sentence, two stances are
clearly being pitted against one another: the stance of “Geraldine Pederson-Krag”
and the opposing stances of her contemporaries. From this brief extract, we again
glimpse the interactional and collaborative nature of stance, as one stance here
is being juxtaposed against another – just as in examples (1)–(3) in the previous
section. The final two principles, indexicality and consequentiality, would require
more than a brief window of textual context. These both necessitate a broader
understanding of the sociocultural and interactional nature of the stances being
taken, which would be virtually impossible to achieve by using collocational evi-
dence alone (cf. Hunston, this volume, in terms of the interplay between quantita-
tive and qualitative analyses).
2.4 Summary
The previous three subsections have offered a quantitative and qualitative over-
view of how speakers and writers use the term stance. Approaching the meaning
of stance from a usage-based perspective recognizes that frequency of use, the
types of language it occurs in, and the broader interactional and collocational
contexts all play a role in how stance is conceptualized.
In terms of frequency of use, Section 2.1 demonstrates that stance is an infre-
quent lexeme, occurs far more often in written discourse than in spoken discourse,
and shows skewed distribution across the sub-genres of language recognized in
the BNC. These distributional facts, especially the lack of a single token of stance
in the 4-million-word conversational component of the BNC, suggest that stance
is a term that tends to occur in quite specialized genre contexts. Even the three
tokens of stance found in the SBCSAE illustrate the restricted contexts of use: one
is a specific reference to posture in a judo class, and the other two are specific to a
conversation about academic institutions – one referring to a college admissions
interview, and the second referring to the institutional views of another school.
It is no wonder, then, that stance has been appropriated for language-related re-
search; this is in keeping with its formally-skewed distribution and prevalence in
specialized genres.
The qualitative analyses of the three tokens in the SBCSAE brings to light
five general principles about stancetaking that can be inferred from the use of the
term stance in the context of interaction; and these same principles in fact will ap-
pear as themes, to varying degrees, in the subsequent papers in this volume. First,
stance refers to physical embodied action, as in example (1), personal belief/at-
titude/evaluation, as in example (2), and/or the social morality espoused at the
institutional level, as in example (3). Secondly, stance is a public act, which is rec-
Introduction 15
ognizable, interpretable, and subject to evaluation by others (cf. Du Bois this vol-
ume). Thirdly, stance is a relational notion (cf. Scheibman this volume); stance is
interactional in nature, collaboratively coming into being among the participants
in an exchange and/or by virtue of opposition to other stances. Fourthly, specific
stances are indexical, evoking larger aspects of the physical context or the socio-
cultural systems in which they are embedded. Finally, stancetaking is consequen-
tial (cf. Du Bois this volume; Keisanen this volume); i.e., taking a stance has real
consequences for the persons or institutions involved. Section 2.3 demonstrates
that the adjectives that tend to collocate with stance provide further support for
these five themes as observed qualitatively in the interactional data.
Our analysis of stance in discourse suggests overwhelmingly that it is by no
means referring to a monolithic concept. Given the diverse nature of stance itself,
as conceptualized in natural discourse, it is no wonder that the representation of
stance in language-related research is also broad and multifaceted. Approaches
to stance from an academic perspective tend to focus to varying degrees on any
number of these five principles, although often this focus is only implicit. For this
reason, stance has been operationalized differently from one researcher to the
next, and a fair amount of work exists that deals with stance-related themes but
that uses other terminology. The following section summarizes a few relevant ap-
proaches to stance from the perspective of language-related research, and serves
to contextualize the present volume within this broader intellectual climate.
3. Research on stance: Subjectivity, evaluation, interaction
As suggested by the subtitle of the present volume – Subjectivity, evaluation, inter-

action – these three terms have each been widely associated with stance. In fact,
most stance literature makes explicit reference to one or more of these categories.
This section provides a cursory overview to contextualize the papers in the pres-
ent volume. This review is not intended to be exhaustive or comprehensive, and
interested readers should of course consult the cited works for more information
and specific details.
Lyons, who has long championed the recognition of subjectivity in language,
defines subjectivity as follows:
We can say that locutionary subjectivity is the locutionary agent’s (the speaker’s
or writer’s, the utterer’s) expression of himself or herself in the act of utterance:
locutionary subjectivity is, quite simply, self-expression in the use of language.
(1994: 13)
Subjectivity, broadly construed, is seen as an essential quality of language. As Ben-

veniste famously speculates: “language is marked so deeply by the expression of
subjectivity that one might ask if it could still function and be called language if
it were constructed otherwise” (1971: 225). However, until the mid 1980s, subjec-
tivity remained out of the scope of interest for most linguists. Lyons (1981: Chap.
10) explores the complex nature of subjectivity and its central role in language
use, and laments that approaches to semantics and pragmatics that fail to fully
embrace subjectivity are inherently flawed. Within the past quarter century after
Lyons’ claim, subjectivity has indeed taken on a more central role in linguistics –
especially in cognitive linguistics and other functionalist paradigms. Langacker
(1985) notes the inherent role of subjectivity in construing a scene and profiling
various aspects of it. The subjective nature of point-of-view and scene construal
is central to cognitive grammar, and has been more fully fleshed out over the
intervening years (Langacker 1999, inter alia). Athanasiadou et al. (2006) pro-
vides a recent and thorough overview of subjectivity from a variety of linguistic
perspectives.
Evaluation has been operationalized by Thompson and Hunston as follows:
evaluation is the broad cover term for the expression of the speaker or writer’s
attitude or stance towards, viewpoint on, or feelings about the entities or propo-
sitions that he or she is talking about. That attitude may relate to certainty or
obligation or desirability or any of a number of other sets of values. (2000: 5)
Thus, evaluation can be roughly summed up as subjectivity with a focus. In oth-

er words, while subjectivity refers broadly to “self-expression” (Lyons 1994: 13),
evaluation implies self-expression that is focused toward a narrow purview – self-
expression about the “entities or propositions” (Thompson and Hunston 2000: 5)
present in the very language that the speaker or writer is currently producing.
Research in evaluation therefore seeks to explicate the range of lexical, grammati-
cal, textual, and intertextual means by which speakers and writers laminate their
language with their attitudes and points-of-view about its content. Hunston and
Thompson (2000) present work on evaluation from a variety of perspectives (see
Thompson and Hunston (2000) for a comprehensive overview). Evaluation figures
prominently both in corpus linguistics and in systemic-functional linguistics. For
the latter, Martin and White (2005) provide a thorough examination of evaluation
from an appraisal framework. Other work on evaluation has focused on elucidat-
ing its features within specific types of language – e.g., Bednarek (2006) provides
a book-length study of evaluation in newspaper discourse.
Both subjectivity and evaluation are strongly implicated in the following two
definitions of stance. Biber et al. define stance as: “personal feelings, attitudes,
value judgments, or assessments” (1999: 966). Similarly, in earlier work, Biber and
Introduction 17
Finegan define stance as “the lexical and grammatical expression of attitudes, feel-
ings, judgments, or commitment concerning the propositional content of a mes-
sage” (1989: 92). Both definitions strongly assert the subjective and evaluative na-
ture of stance. Given the categories proposed in these two definitions, we observe
that stance can be subdivided into evaluation (“value judgments,” “assessments,”
and “attitudes”), affect (“personal feelings”) (cf. Ochs 1989), and epistemicity
(“commitment”). Researching stance, then, according to the two definitions cited
above, entails a clear form/meaning relationship; stance is located in form, i.e., in
“the lexical and grammatical expression” (Biber and Finegan 1989: 92). It is the
goal of the stance researcher, then, according to this approach, to investigate how
lexicon and grammar both encode and reflect the various categories of stance.
Stubbs (1986) has called for a similar research agenda:
whenever speakers (or writers) say anything, they encode their point of view to-
wards it … The expression of such speakers’ attitudes is pervasive in all uses of
language. All sentences encode such a point of view, … and the description of the
markers of such points of view and their meanings should therefore be a central
topic for linguistics. (1986: 1)
Researchers in both corpus linguistics and systemic-functional linguistics have

contributed to the identification and description of lexis and grammar that serve
as markers of stance. In particular, adverbials have proven to be a rich source of
various types of epistemic, attitudinal, and style stances (Biber and Finegan 1988,
1989; Conrad and Biber 2000; Downing 2002). The grammar of English modals
has also been well documented and investigated (cf. review and discussion in
Thompson and Hunston 2000: 20–21), as have evaluative adjectives and nouns
(Hunston and Sinclair 2000). With regard to epistemic stance, or the degree of
commitment/certainty of the speaker/writer, research on complement clauses
and complement-taking predicates has also been fruitful. Field (1997) outlines
the role of factive constructions to index epistemic stance. Kärkkäinen (2003)
similarly approaches epistemic stance by analyzing certain kinds of complement-
taking predicates such as I think – which, following Thompson and Mulac (1991),
have grammaticized into epistemic fragments. See also Kärkkäinen (this volume)
for an analysis of I guess, and Rauniomaa (this volume) for an analysis of two
similar epistemic constructions in Finnish, minun mielestä (‘in my opinion’) and
minusta (‘I think’). Hyland and Tse (2005) provide an overview of what they term
“evaluative that constructions” in academic writing, demonstrating how writers
use these types of complement clauses to display their own stance toward the
information they are presenting.
Other recent studies have focused on stance and subjectivity in naturally-
occurring conversation. Scheibman (2002) provides an overview of frequent
combinations of subjects, verb types, and tenses in American English, arguing

that grammatical and lexical patterns are shaped by subjectivity and the speak-
ers’ needs to personalize their contributions to the discourse. Kärkkäinen (2003),
as already discussed, analyzes subject-verb combinations that serve as epistemic
fragments to index subjectivity and stance in American English conversation. Wu
(2004) analyzes clause-final particles in Mandarin conversation and describes
their use in marking epistemic stance.
Another trend in recent research entails a recognition of, and orientation to,
the sociocultural dimensions of stancetaking. Ochs (1996) writes:
linguistic structures that index epistemic and affective stances are the basic lin-
guistic resources for constructing/realizing social acts and social identities. Epis-
temic and affective stance has, then, an especially privileged role in the constitu-
tion of social life. This role may account in part for why stance is elaborately
encoded in the grammars of many languages. (1996: 420)
Examples of recent work exploring the construction and realization of “social acts
and social identities” (Ochs 1996: 420) through stance include Matoesian, who
provides an analysis of a focus group meeting of police officers and demonstrates
that stance serves to “index broader forms of socio-cultural knowledge embedded
in the professional division of labor between academic trainers and police train-
ees” (2005: 169). Matoesian’s work is also fairly unique among stance research, as
it explicitly addresses the role of physical stance and embodied action. Shoaps
(2004) analyzes Zacapultec ritual wedding councils and explores the various cul-
turally-situated practices of stancetaking. Sheibman (this volume) addresses the
interplay between stance, and the wider social discourses and stereotypes that
those stances may evoke. On a more micro-level of identity, papers by Johnstone
(this volume) and Englebretson (this volume) similarly deal with how speakers
may use stancetaking to index social identities. (See also Bucholtz and Hall (2005)
and Benwell and Stokoe (2006) on identity.)
Other recent approaches have begun to address the cross-cultural, cross-lin-
guistic, historical, and developmental nature of stance marking. Precht (2003)
provides a statistical analysis of stance-related lexemes and grammatical con-
structions in various genres of British and American English, noting cross-cul-
tural differences in the expression of stance. Berman et al. (2002), and papers in
Berman (2005) investigate what they term “discourse stance”:
We consider the notion ‘discourse stance’ as referring to three interrelated di-
mensions of text-construction: Orientation (Sender, Text, Recipient); Attitude
(Epistemic, Deontic, Affective); and Generality (of reference and quantification –
specific vs. general). These are functional dimensions which apply across texts…”
(Berman et al. 2002: 258)
Introduction 19
Scholars working in this approach have focused on analyzing and comparing a

carefully-collected sample of written language from school children, of various
ages and in various countries, to assess the development of “discourse stance”
from a cross-linguistic perspective. Fitzmaurice (2004) presents a historical per-
spective on the development of stance markers in English, illustrating that many
have shifted from a subjective to an intersubjective orientation.
The final trend that I will address in this section concerns the interactional
nature of stancetaking, which takes as a startingpoint the intersubjectivity of lan-
guage. Recent work, especially in cognitive linguistics, has begun to take such a
focus. Verhagen (2005) presents one such notable in-depth study of this topic.
Work on stance, specifically from an intersubjective approach, includes White,
who views stance as an inherently dialogic activity, and investigates the textual
resources that
provide the means for speakers/writers to take a stance towards the various
points-of-view or social positionings being referenced by the text and thereby
to position themselves with respect to the other social subjects who hold those
positions. (2003: 259)
Yet, for researchers whose work is either based in or informed by conversation

analysis (CA), as is the case with many (but not all) of the contributions to the
present volume, intersubjectivity is nothing new. As Edwards (1999) observes,
the theoretical and methodological underpinnings of CA take intersubjectivity
as a given:
In CA, intersubjectivity is not so much a special domain of study, nor something
carried by a special range of devices but, rather, it is a pervasive feature of talk, of
the entire machinery of turn-taking for instance, of uptakes and recipient design,
where every word we use is indexical. Furthermore, intersubjective understand-
ings are part of talk’s business, part of what a stretch of talk is doing or accom-
plishing or proposing. (1999: 131)
Thus, any interactional sequence (such as adjacency pairs, assessment sequences,

and so on) is inherently intersubjective – as implied from the interactional in the
category label interactional sequence. Stance, then, from this perspective, takes its
place among the aspects of language that are jointly constructed, negotiated, and
realized in and through interaction. Heritage and Raymond (2005) provide one
recent study along these lines. Several of the papers in the current volume, espe-
cially the contributions by Keisanen and Haddington, also deal with this aspect
of stance head-on. However, the interactional nature of stance is not just evident
in conversation, but is widely recognized in research on stance in writing as well.
Written texts are broadly understood as interactive and interpersonal, and studies
of stance in writing focus both on how the author engages the readers, as well as
how the readers engage with the text. (Hyland 2005; Hyland and Tse 2005; White
2003; inter alia).
In conclusion, this section has provided a brief overview of trends in stance
research that, broadly construed, include subjectivity, evaluation, and interaction.
This research background provides the intellectual context in which the papers in
this volume have come into being.
4. Conclusion
The previous sections of this introduction have contextualized this volume within
the broader discourse-based and academic notions of stance. The current and
final section briefly introduces each of the contributions to this volume in turn.
Before getting to the specific papers, however, a note on transcription is in order.
Throughout this volume, the reader will notice a variety of transcription systems
and conventions. In the interests of being true to the authors’ intentions, and in
preserving the data as originally transcribed, we have chosen not to pursue a uni-
fied approach to transcription for the volume as a whole. Rather, we have chosen
to allow each author to use the transcription system that best suits his or her
purposes, and to include an appendix of transcription symbols at the end of each
paper as relevant.
Each of the first four papers in the volume (after the introduction) addresses
stance from within a specific approach to linguistics: corpus linguistics, sociolin-
guistics, field linguistics/grammatical description, and discourse linguistics. This
group of papers begins with Hunston’s contribution, a discussion of stance in
large electronic corpora. Hunston demonstrates the essential role of both quan-
titative and qualitative analyses, and shows how these methods can profitably be
used together in the identification and description of stance across and within
various text types. In the next paper, Johnstone offers a sociolinguistic perspec-
tive on stance, exploring the relationship between dialect, local identity, and
epistemic-stance moves. Through an analysis of a sociolinguistic interview with
two residents of Pittsburgh, Johnstone explores the discursive construction of
local identity, and the ways in which speakers use epistemic stance to support
or undermine each others’ claims about local-sounding speech. In the subse-
quent paper, Englebretson presents an overview of three aspects of stancetaking
in colloquial Indonesian, and argues that grammatical descriptions of particu-
lar languages need to pay attention to how grammar functions in the service of
stancetaking. Englebretson presents an overview of three aspects of colloquial
Indonesian grammar (first-person-singular reference, the -nya clitic, and voice
Introduction 21
alternation) and illustrates that in addition to the traditional functions ascribed to

these three grammatical resources, speakers are actively using them to construct
their social worlds through stancetaking. This group of four papers concludes
with Scheibman’s contribution, a study of generalizations in American English
conversation. Scheibman describes the grammatical forms and discursive use
of a set of generalizations in her corpus, demonstrating that generalizations are
both subjective and intersubjective within the context of interactive discourse.
Participants in conversation use generalizations primarily to index solidarity with
one another, and generalizations themselves often index commonly held beliefs
and broader cultural values. Through this analysis of generalizations, Scheibman
highlights the interactive and collaborative nature of stancetaking.
The contribution by Du Bois presents a unified conceptual framework and
model of stancetaking, known as the stance triangle, and serves to some degree
as background for each of the remaining four papers in the volume. Du Bois ar-
gues that stance is a social action that is characterized in terms of the interrelated
dimensions of objective, subjective, and intersubjective orientations. These three
dimensions relate to the act of taking a stance, in which the stance-taker evalu-
ates an object, positions the subject, and aligns with other subjects. These three
facets of stancetaking are mutually constitutive, and, Du Bois shows, are regularly
oriented to in naturally-occurring talk.
Each of the remaining four papers in the volume offers a unique perspective
on stance in talk-in-interaction – informed by Du Bois’s stance triangle model, as
well as by methods of conversation analysis. These four papers are also grouped
together because the authors themselves were members of the stance research
group at the University of Oulu, Finland, under the direction of Elise Kärkkäinen,
and each paper in this group represents significant research based on that project.
In the first paper in this group, Kärkkäinen provides a description and account
of the epistemic fragment I guess in American English conversation. Kärkkäinen
argues that this expression serves to organize stancetaking among co-participants
in talk, and is intimately bound up with the sequential organization and interper-
sonal projections in the conversation. In the following paper, Rauniomaa con-
tinues the theme of epistemic fragments in conversation, providing an analysis
of two stance markers in Finnish, minun mielestä (‘in my opinion’) and minusta
(‘I think’). Rauniomaa observes the role of these epistemic fragments within as-
sessment sequences, and explicates their function in the interactive projection
of sequence organization. In the subsequent paper, Keisanen takes up the theme
of stance in the sequential organization of talk, as she analyzes challenges that
are formatted as negative interrogatives. Keisanen demonstrates the collaborative
nature of stancetaking, in that the act of taking and challenging a stance relies on
at least two participants. She demonstrates the consequential nature of stancetak-
ing, in that once a speaker has taken a stance, this stance is open for challenge by
another participant. The volume concludes with Haddington’s contribution on
stancetaking in news interviews. Haddington observes ways in which interview-
ers position interviewees with regard to stance, and how interviewees in turn dia-
logically align or disalign with that positioning. Haddington shows that stance in
news interviews is both collaborative and consequential.
To conclude, just as stancetaking itself is interactive and emergent in dis-
course, the understanding of stance that emerges out of the papers in this volume
is likewise collaboratively constructed. Each of the authors brings to bear a par-
ticular approach, a certain perspective, and a range of data and questions. Each
contribution highlights key facets of language as used in actual discourse, by real
people, in their full sociocultural environments. Taken together, the papers in this
volume offer a rich overview of this important aspect of sociocultural life and lan-
guage use, and provide a starting point from which to pursue further exploration
and refinement of the burgeoning field of stance research.
Notes
1. Throughout this chapter, italics serve to differentiate mention versus use of the term stance.
When italicized, stance indicates a mention of this English lexical item, in order to focus on it
as a specific term and to scrutinize its meaning and use. When not italicized, no such specific
lexical focus is intended, and it should be understood as a use of this term in the full context of
stance research.
2. All data cited from the BNC in this chapter have been extracted from the British National
Corpus, distributed by Oxford University Computing Services on behalf of the BNC Consor-
tium. All rights in the texts cited are reserved.
3. I wish to gratefully acknowledge Mark Davies for creating such a wonderfully accessible
and user-friendly search tool for the BNC, and for making it publicly available online for lan-
guage research purposes.
4. The figure of 249,000 words for the four volumes of the SBCSAE was arrived at after exclud-
ing non-word tokens such as speaker labels, pauses, and non-vocal noises such as table thumps;
this figure also excludes non-lexical vocal noises such as laughter, coughing, and throat-clear-
ing. This word-count is thus an accurate reflection of the number of spoken words (including
truncated words and so-called filled pauses) in the SBCSAE to date.
5. The figure of 100 million words is generally cited as the size of the BNC. However, the
number of W-Units (tagged words) in this corpus is actually around 97.6 million (officially
97,619,934 W-Units according to http://www.natcorp.ox.ac.uk/corpus/index.xml.ID=numbers,
although this number is slightly less when using the VIEW search interface). Because concor-
dances and queries operate on W-Units, all tables in this section will use the actual word-count
generated by the concordance: 97,619,311 words according to the VIEW search tool.
Introduction 23
6. According to the BNC FAQ at http://www.natcorp.ox.ac.uk/faq.xml. ID=corpus most of

the audio recordings are stored in the National Sound Archive at the British Library where they
can be accessed for research purposes.
7. A reviewer has raised the interesting question as to whether stance is used differently in
American versus British English. Because of the vastly different nature of these two corpora,
this is not a question I can answer here.
8. There is some slight variation in the exact word-count of the written and spoken compo-
nents, depending on whether one uses the official BNC numbers or the numbers arrived at us-
ing VIEW. All numbers in this section refer to those calculated by the VIEW search tool. VIEW
is based on David Lee’s indexing work of the BNC: http://personal.cityu.edu.hk/~davidlee/
devotedtocorpora/home/BNC_WORLD_INDEX.ZIP which claims to have corrected a num-
ber of errors and inconsistencies in the genre classification found in the publication version.
9. The transcription conventions used in the SBCSAE are outlined in Du Bois et al. (1993),
and also appear in this volume as appendices to the papers by Scheibman and Kärkkäinen.
10. Due to space considerations, two of the speaker labels have been shortened from those
presented in the published SBCSAE transcript. The actual speaker labels appearing in the pub-
lished corpus read “STEPHANIE,” “GAIL,” “PATTY,” and “MAUREEN.”
References
Athanasiadou, A., Canakis, C. and Cornillie, B. (eds.). 2006. Subjectification: Various Paths to
Subjectivity. New York: Mouton de Gruyter.
Barlow, M. and Kemmer, S. (eds.). 1999. Usage-Based Models of Language. Stanford: CSLI.
Bednarek, M.A. 2006. Evaluation in Media Discourse: Analysis of a Newspaper Corpus. London:
Continuum International.
Benveniste, E. 1971. “Subjectivity in language.” In Problems in General Linguistics, M.E. Meek
(trans), 223–230. Coral Gables, FL: University of Miami Press.
Benwell, B. and Stokoe, E. 2006. Discourse and Identity. Edinburgh: Edinburgh University
Press.
Berman, R.A., Ragnarsdóttir, H. and Strömqvist, S. 2002. “Discourse stance.” Written Language
and Literacy 5 (2): 255–290.
Berman, R.A. 2005. “Developing discourse stance across adolescence.” Special Issue, Journal of
Pragmatics 37 (2).
Biber, D. and Finegan, E. 1988. “Adverbial stance types in English.” Discourse Processes 11 (1):
1–34.
Biber, D. and Finegan, E. 1989. “Styles of stance in English: Lexical and grammatical marking
of evidentiality and affect.” Text 9 (1): 93–124.
Biber, D., Johansson, S., Leech, G., Conrad, S. and Finegan, E. 1999. The Longman Grammar of
Spoken and Written English. London: Longman.
The British National Corpus, version 2 (BNC World). 2001. Distributed by Oxford University
Computing Services on behalf of the BNC Consortium. URL: http://www.natcorp.ox.ac.
uk/
Bucholtz, M. and Hall, K. 2005. “Identity and interaction: A sociocultural linguistic approach.”
Discourse Studies 7 (4–5): 585–614.
Bybee, J.L. 2006. “From usage to grammar: The mind’s response to repetition.” Language 82 (4):
711–733.
Conrad, S. and Biber, D. 2000. “Adverbial marking of stance in speech and writing.” In Evalua-
tion in Text: Authorial Stance and the Construction of Discourse, S. Hunston and G. Thomp-
son (eds.), 56–73. New York: Oxford University Press.
Davies, M. VIEW (Variation in English Words and Phrases). BNC online search tool. Brigham
Young University: http://view.byu.edu.
Downing, A. 2002. “‘Surely you knew!’: Surely as a marker of evidentiality and stance.” Func-
tions of Language 8 (2): 251–282.
Du Bois, J.W., Chafe, W., Meyer, C. and Thompson, S.A. 2000. Santa Barbara Corpus of Spoken
American English, CD, Part 1. Philadelphia: Linguistic Data Consortium.
Du Bois, J.W., Chafe, W., Meyer, C., Thompson, S.A. and Martey, N. 2003. Santa Barbara Cor-
pus of Spoken American English, DVD, Part 2. Philadelphia: Linguistic Data Consortium.
Du Bois, J.W. and Englebretson, R. 2004. Santa Barbara Corpus of Spoken American English,
DVD, Part 3. Philadelphia: Linguistic Data Consortium.
DVD, Part 4. Philadelphia: Linguistic Data Consortium.
Du Bois, J.W., Schuetze-Coburn, S., Cumming, S. and Paolino, D. 1993. “Outline of discourse
transcription.” In Talking Data: Transcription and Coding in Discourse Research, J.A. Ed-
wards and M.A. Lampert (eds.), 45–89. Hillsdale, NJ: Lawrence Erlbaum Associates.
Edwards, D. 1999. “Shared knowledge as a performative and rhetorical category.” In Pragmatics
in 1998: Selected Papers from the 6th International Pragmatics Conference, vol. 2, J. Ver-
schueren (ed.), 130–141. Antwerp: International Pragmatics Association.
Englebretson, R. (Symposium Organizer). 2004. Stancetaking in Discourse: Subjectivity in In-
teraction. The 10th Biennial Rice Linguistics Symposium, Rice University, Houston, TX,
March 31–April 3.
Field, M. 1997. “The role of factive predicates in the indexicalization of stance: A discourse
perspective.” Journal of Pragmatics 27 (6): 799–814.
Fitzmaurice, S. 2004. “Subjectivity, intersubjectivity and the historical construction of interloc-
utor stance: From stance markers to discourse markers.” Discourse Studies 6 (4): 427–448.
Gardner, R. 2001. When Listeners Talk: Response Tokens and Listener Stance. Amsterdam: John
Benjamins.
Haviland, J.B. 1989. “‘Sure, sure’: Evidence and affect.” Text 9 (1): 27–68.
Heritage, J. and Raymond, G. 2005. “The terms of agreement: Indexing epistemic authority and
subordination in talk-in-interaction.” Social Psychology Quarterly 68 (1): 15–38.
Hyland, K. 2005. “Stance and engagement: A model of interaction in academic discourse.” Dis-
course Studies 7 (2): 173–192.
Hyland, K. and Tse, P. 2005. “Evaluative that constructions: Signalling stance in research ab-
stracts.” Functions of Language 12 (1): 39–63.
Hunston, S. and Sinclair, J. 2000. “A local grammar of evaluation.” In Evaluation in Text: Au-
thorial Stance and The Construction of Discourse, S. Hunston and G. Thompson (eds.),
74–101. New York: Oxford University Press.
Hunston, S. and Thompson, G. 2000. Evaluation in Text: Authorial Stance and The Construction
of Discourse. New York: Oxford University Press.
Jaffe, A. (Panel Organizer). 2004. Stance in Social and Cultural Context. Special Session at the
Sociolinguistics Symposium 15, Newcastle upon Tyne, April 1–4.
Introduction 25
Kärkkäinen, E. 2003. Epistemic Stance in English Conversation: A Description of Its Interactional

Functions, with a Focus on ‘I think’. Amsterdam: John Benjamins.
Langacker, R.W. 1985. “Observations and speculations on subjectivity.” In Iconicity in Syntax, J.
Haiman (ed.), 109–150. Amsterdam: John Benjamins.
Langacker, R.W. 1999. Grammar and Conceptualization. New York: Mouton de Gruyter.
Lyons, J. 1981. Language, Meaning and Context. London: Fontana Paperbacks.
Lyons, J. 1994. “Subjecthood and subjectivity.” In Subjecthood and Subjectivity: Proceedings of
the Colloquium ‘The Status of the Subject in Linguistic Theory’, M. Yaguello (ed.), 9–17.
Paris: Ophrys.
Macken-Horarik, M. and Martin, J.R. (eds.). 2003. Negotiating heteroglossia: Social perspectives
on evaluation. Special Issue, Text 23.
Martin, J.R. and White, P.R.R. 2005. The Language of Evaluation: Appraisal in English. New
York: Palgrave Macmillan.
Matoesian, G. 2005. “Struck by speech revisited: Embodied stance in jurisdictional discourse.”
Journal of Sociolinguistics 9 (2): 167–193.
Mushin, I. 2001. Evidentiality and Epistemological Stance: Narrative Retelling. Amsterdam: John
Benjamins.
Ochs, E. (ed.). 1989. “The pragmatics of affect.” Special Issue, Text 9 (1).
Ochs, E. 1996. “Linguistic resources for socializing humanity.” In Rethinking Linguistic Relativ-
ity, J. Gumperz and S. Levinson (eds.), 407–437. New York: Cambridge University Press.
Precht, K. 2003. “Stance moods in spoken English: Evidentiality and affect in British and Amer-
ican conversation.” Text 23 (2): 239–257.
Scheibman, J. 2002. Point of View and Grammar: Structural Patterns of Subjectivity in American
English Conversation. Amsterdam: John Benjamins.
Shoaps, R.A. 2004. Morality in Grammar and Discourse: Stance-taking and the Negotiation of
Moral Personhood in Sakapultek (Mayan) Wedding Counsels. Ph.D. dissertation. Linguis-
tics Department, University of California, Santa Barbara.
Shoaps, R. and Kockelman, P. (Panel Organizers). 2002. Morality and Epistemology: Stance-
taking and the Discursive Constitution of Personhood. Panel presented at the 101st annual
meeting of the American Anthropological Association, New Orleans, LA, November 21.
Silverstein, M. 1976. “Shifters, verbal categories, and cultural description.” In Meaning in An-
thropology, K.H. Basso and H. Selby (eds.), 11–55. Albuquerque: University of New Mex-
ico Press.
Stubbs, M. 1986. “A matter of prolonged fieldwork: Notes towards a modal grammar of Eng-
lish.” Applied Linguistics 7: 1–25.
Thompson, G. and Hunston, S. 2000. “Evaluation: An introduction.” In Evaluation in Text:
Authorial Stance and the Construction of Discourse, S. Hunston and G. Thompson (eds.),
Thompson, S. and Mulac, A. 1991. “A quantitative perspective on the grammaticization of epis-
temic parentheticals in English.” In Approaches to Grammaticalization, vol. 2, E.C. Trau-
gott and B. Heine (eds.), 313–329. Amsterdam: John Benjamins.
Verhagen, A. 2005. Constructions of Intersubjectivity: Discourse,Syntax, and Cognition. New
York: Oxford University Press.
White, P.R.R. 2003. “Beyond modality and hedging: A dialogic view of the language of inter-
subjective stance.” Text 23 (2): 259–284.
Wu, R.-J. 2004. Stance in Talk: A Conversation Analysis of Mandarin Final Particles. Amsterdam:
John Benjamins.
Using a corpus to investigate stance
quantitatively and qualitatively
Susan Hunston
University of Birmingham
1. Introduction
Since their development in the 1960s, electronic collections of large numbers of

texts, known as corpora, have been used in increasingly diverse ways to investigate
topics such as lexis, grammar, stylistics, language change, and language acquisi-
tion (Biber et al. 1999; Granger 1998; Lindquist and Mair 2004; Semino and Short
2004; Sinclair 1991). Standard corpus investigation software allows the researcher
to examine the frequency of words, phrases, or structures in a given corpus and
to calculate comparative frequencies of co-occurrence. Such calculations have led
to, among other things, an increased concern for the concept of collocation (Sin-
clair 1991, 2004) and for the examination of variation between registers (Biber
1988; Matthiessen 2006). In addition, software allows the researcher to look at
each example of a given word or phrase in its immediate context in a collection
of texts. Such investigations reveal the “latent patterning” of language (Sinclair
and Coulthard 1975: 125), showing the importance of context and intertextuality
to meaning (Teubert 2004), and suggesting that meaning is disseminated across
phraseologies rather than belonging to individual words (Sinclair 1991, 2004).1
There are two complementary aspects of corpus linguistics that are relevant
to this paper, both of which demonstrate the benefits of examining not just natu-
rally-occurring data, but large amounts of such data. Firstly, corpora give us the
opportunity to quantify and make it particularly easy to quantify forms. Secondly,
they allow us to observe multiple uses of a word or phrase in context.
The phenomenon of stance represents an area of difficulty for corpus linguis-
tics, because stance is a meaning, a type of meaning, or several types of mean-
ing, rather than a form. It is true that papers in this collection and elsewhere
identify particular forms associated with stance in various languages, but it is
28 Susan Hunston
always acknowledged that identifying stance entails more than simply locating
those forms, and that interpreting the role of stance in discourse entails a deeper
understanding of the discourse as a whole than can be obtained from looking at
the immediate co-text of an individual lexical item.
These problems notwithstanding, I shall argue below that corpus methods
can make a useful contribution to the investigation of stance, by drawing atten-
tion to a number of studies in this area. In doing so, I focus on the methodological
issues involved, and I make two assumptions. The first is that quantitative meth-
ods are not irrelevant to discourse studies, in the sense that recurring instances
of a phenomenon are noted, the explication of a single instance normally implies
that a pattern has been identified, and the explanation would hold true for other
similar instances. This is the case even when the amount of data collected is rela-
tively small; quantitative does not mean huge, but simply that statements of the
type ‘this is a demonstrably typical occurrence’ are worth making. The second
assumption is that, on the other hand, research in the area of discourse will never
be wholly quantitative. In fact, the numbers themselves are derived from close ex-
amination of many or all of the specific instances of the targeted phenomenon. In
addition, of course, numbers themselves are useless without interpretation, and
without the grounds for the interpretation being made clear.
The paper begins with two examples of largely quantitative studies of stance,
both of which compare the frequency of stance exponents in two or more collec-
tions of texts. Conrad and Biber examine stance adverbials in four broadly-de-
fined registers (see Section 2), while Charles compares stance-related patterns in
two academic disciplines (see Section 3). The issues that such quantitative work
raises are discussed in Section 4. The paper then turns to issues of qualitative
research and the role of a corpus in that research. Sections 5 and 6 illustrate the
use of a corpus to explain subjective responses to individual texts, in particular
to find an account for the sometimes hazy intuitive sense that an utterance has a
stance implication even when there are no obviously evaluative items in it. It is
argued that a corpus can help to demonstrate that many items, though not carry-
ing stance themselves, often co-occur with stance items. This argument is carried
forward in Section 7, where it is proposed that the combination of such features
amounts to a rhetoric of stance. Section 8 concludes that, although a comprehen-
sive account of stance cannot be wholly quantitative, the availability of quantita-
tive data in a corpus can assist the investigation of stance in texts in both qualita-
tive and quantitative ways.
Using a corpus to investigate stance 29
2. An example: Stance adverbials
One of the best-known corpus studies of stance is Conrad and Biber’s (2000)
study of stance adverbials, which they define as “grammatical devices used to
frame a proposition” (Conrad and Biber 2000: 58). They divide these adverbials
into a number of cross-cutting categories:
• Meaning. Conrad and Biber distinguish epistemic stance (e.g., evidently), at-
titudinal stance (e.g., most surprisingly), and style stance (e.g., simply put).
• Form. They distinguish between single adverbs (evidently), adverb phrases
(most surprisingly of all), noun phrases (no doubt), prepositional phrases (for
a fact), finite clauses (I think), and non-finite clauses (more simply put).
• Position. They identify four positions for stance adverbials: initial, pre-verbal,
post-verbal, and final.
Conrad and Biber proceed to compare three broad registers: conversation, ac-
ademic prose, and news reportage with reference to these three variables, by
identifying and quantifying the occurrence of instances of each variable in cor-
pora representing those registers. This exercise allows them (Conrad and Biber
2000: 63–72) to draw a number of conclusions, such as:
• stance marking is particularly frequent in conversation;

• marking of epistemic stance is more frequent overall than marking of attitu-
dinal or style stance;
• single adverbs are more frequent than the other grammatical realizations in
all registers;
• prepositional phrases are more frequent in academic prose than in the other
registers;
• finite clauses are more frequent in conversation than in the other registers;
• final position adverbials are frequent in conversation;
• pre-verbal and initial adverbials are frequent in the written registers.
As well as these general conclusions, they are also able to examine each group of
adverbials in more detail, showing which adverbials are most frequent overall,
and giving information about the uses of particular meanings, forms, and posi-
tions. For example, they note that style adverbials occur in news reportage “in
articles which review sports or entertainment performances, and in quotations”
(Conrad and Biber 2000: 67); that “comment clauses” such as I think, I guess, and
I bet account for most of the finite-clause adverbials in conversation (Conrad
and Biber 2000: 70); and that style adverbials in initial position comment both
30 Susan Hunston
on what is coming up and on what has gone before, thereby linking two clauses
(Conrad and Biber 2000: 71).
Conrad and Biber’s work is broad in scope but limited in the amount of detail
it incorporates. It examines only one kind of stance marker and establishes a very
limited number of categories of stance meaning, thereby conflating a large num-
ber of different meanings in a single grouping. Most seriously, perhaps, it uses an
extremely broad definition of register, not distinguishing, for example, between
different types of academic prose or between different contexts of conversation.
Their work is a good example of one kind of corpus study, one that delimits and
counts categories, and that compares corpora built according to different criteria.
It is naturally open to criticism in terms of the natural fuzziness of categories of
this kind, and in terms of the specificity of the corpora concerned. Bearing these
criticisms in mind, however, their focus on the variability between registers has
inspired more detailed corpus studies that look at stance features beyond adverbi-
als and break down the broad register categories that they use.
3. Pattern and stance
One example of work that is broadly in the tradition of Conrad and Biber, but
that moves beyond it methodologically, is Charles (2004). Charles restricts her
corpora to two discipline areas: political science and materials science, and to a
particular kind of writing: postgraduate student theses. She investigates a number
of classes of items that realize stance: stance adverbials, adjectives in patterns be-
ginning with anticipatory it, nouns that occur both with appositive that-clauses
and as anaphoric nouns (Francis 1994), and report verbs.
Charles’s work draws on the notions of grammar pattern and of meaning
group, as developed by Francis, Hunston, and Manning (e.g., Francis et al. 1996,
1998; Hunston and Francis 1999). Put briefly, pattern identifies the frequent be-
haviour of a given lexical item, expressed as a sequence of elements. Those iden-
tified by Francis et al. include some traditional complementation patterns such
as V ‘that’ (verb complemented by that-clause e.g., ‘believes that’) or V n n (verb
complemented by two noun groups e.g., ‘call him a liar’ or ‘give her a book’), but
also some that are less traditional, such as V n ‘into’ -ing (verb followed by noun
group and prepositional phrase introduced by into e.g., ‘…talk him into going…’),
or N ‘at’ n (noun followed by prepositional phrase introduced by at e.g., ‘…his
anger at the killings…’), or ADJ ‘to’ n (adjective followed by prepositional phrase
introduced by to e.g., ‘…oblivious to the consequences…’).
One of the key findings in this work is that words sharing a pattern can be
grouped according to sense. For example, the pattern V ‘on’ n has a number of
meaning groups including:
1. Verbs meaning ‘think’: brood, cogitate, deliberate, meditate, muse, ponder, re-
flect, ruminate, speculate
2. Verbs meaning ‘add details’: elaborate, embroider, enlarge, expand
3. Verbs meaning ‘have an effect on’: act, impact, impinge, press, tell, weigh
4. Verbs meaning ‘interrupt’: encroach, infringe, intrude, trespass
5. Verbs meaning that someone changes their mind: back-pedal, backtrack, com-
promise, default, prevaricate, rat, renege
6. Verbs meaning ‘depend’: bank, count, depend, hang, hinge, lean, pivot, rely,
rest, ride, turn
7. Verbs meaning ‘spy’ or ‘inform’: eavesdrop, snoop, spy, listen in, grass, inform,
rat, snitch, tell
Some patterns have been identified from previous research (e.g. Francis et al. 1998)
as tending to occur predominantly in the context of the expression of stance, and
these are the ones targeted by Charles as a useful starting point for searching a
relatively large corpus. They include:
• it v-link ADJ that (it’s possible that, it’s amazing that, etc.)
• it v-link ADJ to-inf (it’s easy to see, it’s possible to imagine, etc.)
• N that (the assumption that, the suggestion that, etc.)
• V that (suggests that, believes that, etc.)
For each of these patterns, meaning groups are reported in Frances et al. (1996,
1998). Charles uses the same groups where possible, making additions or correc-
tions where necessary. As an example of such groups, nouns with the pattern N
‘that’ are sub-divided into:
• The idea group: idea, assumption, belief, conclusion, hypothesis, notion, expec-
tation, impression, etc.
• The argument group: argument, assertion, point, statement, comment, sugges-
tion, etc.
• The evidence group: evidence, indication, observation, proof, etc.
• The possibility group: possibility, chance, danger, etc.
There are a number of advantages to this use of pattern. Although the instances of
stance identified are still limited by form – Charles, like Conrad and Biber, makes
no claim to be exhaustive in her quantification of stance – these instances comprise
words in context rather than in isolation. Whereas the word possible, for example,
32 Susan Hunston
may have many different functions in a text, the functions of it is possible that
or it is possible to are much more limited. Furthermore, using patterns such as N
‘that’ and ‘it’ v-link ADJ ‘that’ allows Charles to incorporate nouns and adjectives
into her study, yet still to keep the study within manageable bounds. The use of
meaning groups allows her to identify more specific types of meaning than Conrad
and Biber do, while at the same time grouping together, and so taking account of,
numbers of words each of which may occur only infrequently. For example, in the
argument group of nouns occurring in the pattern N ‘that’, there are 36 nouns in
Charles’s politics corpus, many occurring only once or twice, and 9 in her materials
corpus, of which 7 occur once or twice. Counting these nouns together allows even
the very infrequent ones to be included in the quantification.
Charles’s quantification of instances of stance, using this method, shows dis-
tinctions between the disciplines under investigation. She notes, for example, that
nouns in the pattern N ‘that’ and in the argument group occur just over 45 times
per ten-thousand words in the politics corpus compared with just over 5 times
per ten-thousand words in the materials corpus. By comparison, nouns in the
evidence group occur with approximately equal frequency, as do nouns in the
possibility group. Charles’s figures for these four groups are shown in more detail
in Table 1.
Charles acknowledges, however, that figures such as these give only general
information, and she goes on to make more interesting observations about the
differences between the two corpora, moving from the quantitative to the qualita-
tive. There is space here only to give a few examples of her work.
In her study of adverbs, one of the meaning groups Charles identifies is the
generally group, consisting of generally, as a whole, usually, in general, and typi-
cally. These are about twice as frequent in the materials corpus as in the politics
corpus. She also identifies a particularly group (particularly and in particular),
which is more frequent in the politics corpus than in the materials corpus. Cit-
ing Becher and Trowler (2001), Charles links this difference to a contrast in what
counts as knowledge in the two disciplines: “knowledge in the natural sciences
focuses primarily on universals, while that in the social sciences is more likely to
be concerned with particulars” (Charles 2004: 42). However she also notes more
Table 1. Meaning group frequency (adapted from Charles 2004: 118)

Group Politics corpus Materials corpus
(tokens per 10,000 words) (tokens per 10,000 words)
Idea 72.6 13.7
Argument 45.3 5.7
Evidence 14.2 15.0
Possibility 6.8 5.7
subtle differences, such as the use of the individual word generally in the two dis-
ciplines. The main use in materials is to make a general statement within which
the current writer locates their work; whereas the main use in politics is to make
a general statement against which the current writer locates their work. Charles
(2004: 43) quotes these examples in illustration:
(1) It is difficult to calculate this force and it is generally taken into account using
a statistical treatment… We make the same assumption for a single particle
(MS)
(2) Thus far I have concentrated on the factors generally cited as having been
responsible for the decision to abandon Britain’s world role. But they have
been found to be insufficient to explain the decision. (PS)
As noted above, Charles divides nouns in the pattern N ‘that’ into meaning groups
and compares their frequency. In addition, she notes that nouns in the idea group
are used in the politics corpus to comment on the work of others; but in the ma-
terials corpus to refer to the writer’s own work. Examples from her study include
(Charles 2004: 120–121):
(3) The difficulty with the assumption that ‘middlepowermanship’ was a function
of ‘middlepowerness’ was that it was demonstrably untrue (PS)
(4) These results are consistent with the idea that the smaller particles not only
require less time to oxidise… (MS)
Charles notes that the pattern ‘it’ v-link ADJ ‘that’ occurs with similar frequency
in both corpora. In the politics corpus, however, it is used to comment, not on
research, but on the actions of political actors. In the materials corpus, it is used
to comment on actions undertaken by the thesis writer. In both cases, arguably,
the pattern is used to comment on the object of research. Examples are (Charles
2004: 95):
(5) It quickly became clear, however, that the Bosnian Serbs were not eager to
settle the conflict. (PS)
(6) On removal it was apparent that some of the components of the fibreboard
had melted. (MS)
This pattern, ‘it’ v-link ADJ ‘that’, has proved useful in distinguishing between
other corpora. Hunston (2004), using Charles’s methodology, compares the oc-
currence of the same pattern in two corpora: one consisting of issues of the New
Scientist magazine, the other of issues of the Sun and News of the World news-
papers (popular British tabloids).2 Although the most frequent adjectives in the
34 Susan Hunston
pattern are largely the same in both corpora (9 of the most frequent 12 adjec-
tives in the pattern in each corpus are the same), looking at more instances and a
wider range of adjectives divided into meaning groups indicates that in the New
Scientist, the group including possible and likely is most significant in occurrence,
whereas in the Sun/News of the World, the group including important and vital,
and the group including outrageous, sad, disgraceful, disgusting, disappointing,
and scandalous are most significant. In other words, while the New Scientist com-
ments soberly on the likelihood of a proposition, the tabloid newspapers make a
judgment as to its desirability. The difference is illustrated by these examples:
(7) It was important to establish this because it was possible that strontium and
calcium in fossils might have reacted chemically with the rock in which the
fossils were buried. (New Scientist)
(8) It is scandalous that the rich can buy the drugs privately, but tough luck if you
are poor. (Sun)
Before continuing to discuss the problems associated with quantitative methods,

it is worth pausing to consider what ‘quantitative methods’ of the kind outlined
above involve. In selecting a pattern such as ‘it’ v-link ADJ ‘that’ for study, Charles
and Hunston have focused on a phraseology that is relatively stable, in the sense
that it can be demonstrated, by looking at each instance of use if necessary, that
the overwhelming majority of its instances do represent stance. This does not
mean that each instance is exactly the same as the others, but it does suggest that
broad generalisations can be made. Distinguishing between adjectives that occur
frequently in that pattern in different types of text, and counting the occurrence
of each adjective, again allows generalisations to be made without saying anything
about the specific instances of use. It will be noted that such generalisations are
seldom, even in quantitative approaches, taken as the end point of the investiga-
tion. Charles’s findings (outlined above) concerning the difference between her
Political Science texts and her Materials Science ones in the typical use of word
and phrases such as generally and idea that are the outcome of a close examina-
tion of each instance, often involving the scrutiny of several paragraphs or pages
of text, rather than of a simple counting of words. On the other hand, her concern
is the quantitative one of establishing what is usually or typically the case in each
discipline.
4. The problems of quantification
In spite of examples such as the above, it will be apparent that quantifying stance
is problematic because there is no simple correspondence between individual
words, on the one hand, and stance functions, on the other. As a result, attempts
to quantify stance, especially evaluative stance, by counting particular words are
likely to be unsuccessful. This is exemplified by Groom (2004), who discusses
evaluative language in academic disciplines.
Groom’s topic is the adjectives used in book reviews in academic journals of
history and of English literature. He takes as his starting point Becher’s (1989)
research establishing groups of adjectives – which according to interviewed re-
searchers, have an evaluative meaning in various disciplines – in each case, re-
flecting the value-system of the discipline. For example, Becher says that in his-
tory, positive values are associated with the concepts of scholarly, original, and
well-written, whereas negative values are associated with thin and trivialising. Av-
erage work in history, according to Becher’s findings, might be described as sound
or well-researched. In physics, on the other hand, elegant and economical describe
good work, while poor work is sloppy, and average work is accurate.
One possible conclusion from this research is that it should be possible to
measure the quantity of stance marking in a given text or corpus by counting
the instances of these adjectives and others that are similar. In his corpus of his-
tory review articles, Groom does find large numbers of the adjectives noted by
Becher, but looking more closely, he finds that they are by no means always used
to indicate stance. For example, scholarly is indeed used positively in the following
examples:
(9) …this is a fine, scholarly, and relatively short book….

(10) Siegel has written an informative, scholarly study that is a delight to read…
but is unrelated to stance in many more examples, where the meaning might be
glossed as ‘pertaining to scholars’ rather than ‘demonstrating good scholarship’:
(11) This book features the normal scholarly apparatus of a bibliography…

(12) There will always be scholarly concern over…
(13) …from the viewpoint of provoking scholarly debate…
Overall, Groom interprets only 48 of the 629 instances of scholarly in his corpus
as having a positive evaluative meaning. Similarly he looks at the word original,
which occurs 648 times and which is sometimes positive, as in:
36 Susan Hunston
(14) …this volume comprises a detailed work of original research embellished

with…
but not necessarily so:
(15) …his discussion is not based on original research but instead a synthesis of
selected secondary sources… (lack of originality is not a problem)
(16) Michael Farry’s study of County Sligo may not rival the sweep of Fitzpatrick’s
analysis, and certainly not the pungency of his writing, but it provides a work-
manlike slab of original research on a vital moment in modern Irish history
(originality is not enough)
Again, Groom suggests that only 258 of the 648 instances actually indicate posi-
tive stance.
In other words, Groom argues that immediate co-text distinguishes evalu-
ative from non-evaluative meanings, and that this must therefore be taken into
account in quantifying stance. Looking at each instance of a word in a corpus has
its own problems, however. It militates against the use of the large corpora that are
essential for reliable quantitative results, and it presupposes that the categories of
stance are uncontroversially distinguished.
5. Corpora and the qualitative investigation of stance
It has been suggested so far in this paper that a corpus may be useful in quanti-
fying stance markers, and that this quantification may lead on to more detailed
qualitative work. The point has also been made, however, that context is crucial
in identifying stance, and that this must be borne in mind when relatively crude
quantitative measures are used. Put simply, because there is no one-to-one cor-
relation between form and function, counting forms is not the same as counting
functions. On the other hand, there are relations between form and function, and
the relations become closer the more specific the form is taken to be (e.g., if it is
a phrase, or a word in a pattern, rather than a word). In some instances, then,
counting forms can be a useful first step in quantifying functions.
The emphasis on context leads to an aspect of corpus work that is more
qualitative than quantitative, although it has the potential to lead back into the
quantitative. This is the tradition of examining the typical contexts (co-texts) of
individual words and phrases in order to identify their phraseologies and their
functions. This tradition is beneficial to the study of stance.
A good example is Channell’s (2000) study of the phrase par for the course.
Channell points out that looking at a large number of examples of this phrase
enables the researcher to identify a common context and a common function

of the phrase. She notes that, in its non-golfing uses, par for the course is used to
comment on two aspects of a situation: it is non-remarkable, and it is negatively
evaluated. Examples from Channell’s paper (2000: 47) include:
(17) …the vicious infighting that was par for the course…
(18) The third was out of order, which was par for the course…
(19) A for sexuality and E for subtlety. Par for the course out here…
(20) …was cancelled at the last minute: par for the course…
(21) Disturbing dreams are par for the course in pregnancy…
Channell also points out that in dialogue the phrase is used frequently by a second
speaker to show sympathy for the first speaker. She quotes (2000: 48) the follow-
ing example, in which speaker F is complaining about problems in printing a
computer file:
(22) F: …Soon as you hit ‘Okay’ that’s it. Terminal. So erm I can only think
there’s something in the file that’s doing it.
G: Oh shit. Well it’s par for the course for today I think.
In this example, par for the course indicates an attitudinal stance and also performs
a further interpersonal function in implying solidarity – expressing a shared ac-
ceptance that life is going to deal us blows that we simply have to weather.3
The notion of typicality (Sinclair 1991) is crucial to this kind of corpus use
and has been used to argue for “hidden meanings” in those cases where a phrase
is used in an atypical way (Louw 1993). One example is the following sentence,
which comes from an email message: I’m afraid I may have given Martin the
wrong end of the stick. It can be shown that the phrase the wrong end of the stick
is being used atypically here. The more typical use is illustrated by the following
randomly-selected concordance lines, which have been obtained by searching for
the string wrong end of the stick:4
realises that he got completely the wrong end of the stick, and is
women somehow managed to get the wrong end of the stick. Women
treatment, but sometimes they get the wrong end of the stick. I re
in the head, the public gets the wrong end of the stick, is ign
at cross purposes -- getting the wrong end of the stick -- goin
the dail. ‘He must have got the wrong end of the stick’ said o
that is because they’ve got the wrong end of the stick and the
calls back erm I think she’s got the wrong end of the stick and blo
<ZF1> you’ve <ZF0> you’ve got the wrong end of the stick there.
38 Susan Hunston
too many men have got the wrong end of the stick. They m
4 February 1995) has the wrong end of the stick as far
him off.” STIMAC then got hold of the wrong end of the stick and tho
give wrong advice or get hold of the wrong end of the stick. There
applicant obviously got hold of the wrong end of the stick.” Actua
And she’d really got hold of the wrong end of the stick, becaus
In most of these instances, the phrase the wrong end of the stick follows get or get
hold of. I would argue that in each case responsibility for the misunderstanding
is represented as lying at the door of the receiver of the message, not the giver of
it. This argument was based first on an intuitive interpretation of the examples; as
not all speakers share that intuition, however, further evidence is necessary.5 The
problem is that, while get hold of implies agency, get can be used either to imply
agency (as in I’m going to get a drink) or with a meaning similar to ‘be given’ (as
in it was a big surprise to get the Young Player award). Thus, an example such as
the public gets the wrong end of the stick could mean either ‘the public misun-
derstands’ or ‘someone misleads the public.’ Examination of more context in 25
examples (one-third of the total in the Bank of English) only partially resolves the
issue. In fact, most instances of get the wrong end of the stick do have something in
the surrounding co-text that clearly ascribes blame to the holder of ‘the stick.’ Ex-
amples include the gaffe-prone Prince had got the wrong end of the stick again, and
It is, however, very easy for an individual to establish a limited mind-set and get the
wrong end of the stick. (It might be noted in passing that this search for evidence
goes well beyond the confines of the standard 80-character concordance line, into
the whole text if necessary.) It is true, however, that other examples have no such
contributing evidence and so are potentially ambiguous (e.g., Q: And, lastly, is the
Thumbs Up control of the recording VCR solely by infrared remote control? A: You
seem to have got the wrong end of the stick.) If the search is expanded to less meta-
phoric phrases such as the wrong impression or the wrong idea, however, an inter-
esting asymmetry emerges. The phrase the wrong impression, for example, occurs
140 times, of which 72 occur with the verb give and 33 with get. This suggests that
either speaker or hearer can be deemed to be responsible for the wrong impres-
sion. Frequently, where the verb is get, agency is at least ambiguous, as in If you let
sloppy errors slip by, your teacher will get the wrong impression. In the case of the
wrong end of the stick, however, there are no co-occurrences with give. In other
words, with this phrase, clear ascription of agency to the speaker never or very
rarely occurs. This negative evidence provides some support for the view that with
this phrase the hearer is typically construed as responsible for the misunderstand-
ing, so that a clear indication of the contrary, as in the cited email, is markedly
atypical. Following Louw’s suggestion that atypical usage can indicate insincerity,
using the phrase with give instead of get might indicate that what appears to be an
apology for causing a misunderstanding is actually an accusation.
Sinclair uses the term semantic prosody to refer to instances where an extend-
ed phrase, or “unit of meaning,” such as get the wrong end of the stick has a con-
sistent pragmatic or discourse function, typically involving attitudinal meaning.
Often, the semantic prosody of a phrase is difficult to intuit and can be observed
only by using corpus search techniques. Such phrases are, however, important
embodiments of stance.
6. Where is stance located?
Sinclair’s work highlights the fact that evaluative meaning does not occur in
discrete items but can be identified across whole phrases, or units of meaning,
and that it is cumulative. This is partly what lends stance its subtlety in text: it is
difficult to pin down. Hunston (2004) discusses this problem in relation to the
opening paragraph of an article about stress among English-language teachers,
reproduced here:
(23) The intention of this article is…to share a very real concern about our survival
both as teachers and as human beings. If this seems unduly dramatic, then
so be it. Our situation is already dramatic almost to the point of tragedy. We
are at risk – from the pressures of consumerism, the media, technology, and
rampant trivialisation. Our lives are lived at an increasingly accelerating pace,
leaving less and less time for mature reflection and the exercise of indepen-
dent choice. (Maley 1999)
This paragraph is unapologetically opinionated and highly evaluative. Its purpose

is to tell the reader that something is wrong. Hunston (2004) argues, however,
that it is difficult to be specific about where in the paragraph this stance is ar-
ticulated. More particularly, is the stance to be found in the words and phrases of
the paragraph: a very real concern about, tragedy, at risk, the pressures of, or is it
something that the reader brings to the paragraph with his/her experience of life,
interpreting as undesirable such concepts as living at a fast pace, not having time
for reflection and choice?
There are in fact several different answers to this question. First, there are
words here, which, unless they are negated or attributed, indicate stance on the
part of the writer, such as tragedy.6 Second, there are words that indicate different
kinds of stance, depending on context, such as dramatic. Third, there is phraseol-
ogy, which is frequently associated with stance even though it does not appear to
40 Susan Hunston
indicate stance in itself, such as to the point of. Finally, there are examples that are
more subtle still, where considerable corpus investigation and interpretation is
needed to make sense of the evidence. The example to be illustrated below is an
increasingly accelerating pace. Each of these examples will now be considered in
greater detail (following Hunston 2004).
6.1 Tragedy
A search of the Bank of English corpus suggests that tragedy is typically used as a
count noun and refers anaphorically or exophorically to an event assumed to be
recognised by the reader. One example recalls a shipping accident and recounts
the results of a government inquiry, continuing:
(24) Survivors of the tragedy have been angered by the report’s findings.
A search for of tragedy yields non-count uses, all of which, except those refer-
ring to a dramatic genre, indicate a more general sense of misfortune and sorrow.
Typical examples include:
(25) …my explanation of recent inner-city history is filled with a deep sadness and
sense of tragedy…
(26) Many of the graves tell their own moving stories of a night of tragedy at sea.
(27) I am sometimes asked why I believe, when there is so much evidence of trag-
edy and evildoing.
Apart from the count/non-count distinction, which separates the specific from
the general, there is little apparent variation in the use of the word. Its appearance
in example (23) indicates evaluative stance.
6.2 Dramatic
Investigation of this word in the Bank of English corpus suggests that, among
its other uses, it can attribute to a speech act the quality of ‘unreliable exaggera-
tion’ or it can indicate a judgment of an action as ‘a significant activity with nega-
tive effects.’ In example (23), both meanings appear. First, the writer ascribes to
the reader the sense that the writer’s statements are unduly dramatic (unreliable
and exaggerated); second, he asserts that the situation is indeed dramatic to the
point of tragedy (significant, with negative effects). The two possible meanings of
the word are exploited to embody two competing stances: that of the imagined
Table 2. Words preceding and following to the point of

close to the point of being
thin absurdity
naïve self
simple idiocy
modest paranoia
cautious invisibility
familiar recklessness
quiet farce
misleading emaciation
plain stupidity
intense rudeness
frank near
brave obsession
polite madness
sincere lunacy
charming insanity
confident caution
cool non
shy making
careful cruelty
blunt vulgarity
selfish impossibility
curious blandness
reader, and that of the writer himself. The stances are conveyed in each case by
the word in context.
6.3 To the point of
Although to the point of does not in itself indicate stance, it can be shown to co-
occur, in almost all instances of its use in the Bank of English, with what might
simplistically be called negative evaluation. More accurately, the phrase is used
to indicate that a normally acceptable attribute or behaviour is exaggerated in a
particular instance so that it becomes much less acceptable. Table 2 lists the words
occurring to the left and right of to the point of, in order of frequency, while the
concordance lines show specific examples.
essures which saw him tense to the point of obsession over every to
Restaurants are often basic to the point of being downright scruffy
ll copy them slavishly, even to the point of looking ridiculous. If
42 Susan Hunston
were also low and restricted to the point of inadequacy. The navigat
we thought he was obsequious to the point of irritation, but later r
nfused; one minute laid-back to the point of torpor, the next a mass
all enjoyed the joke almost to the point of seizure, so you will re
ubject of trade, it is vague to the point of being coy. The two side
can be close-mouthed almost to the point of curtness. Ten days ago
articulate and high-handed to the point of arrogance, told the imp
efers to his cinematographer to the point of throwing away necessary
In not every case, however, does the word to the right of to the point of clearly in-
dicate a negative evaluation. Apparent counter-examples include the following:
(28) At 23, and with just one exhibition behind her, Hazel Dooney is fresh to the
point of invigoration.
(29) This assessment, given by one of the nine members of his committee, reflects
a genuine warmth and respect for their chairman. Unfailingly courteous, to
the point of gentleness, Lord Nolan also displayed a fierce independence and
skill in both bringing together his disparate committee members…
The second of these demonstrates how well the phrase to the point of does its job.
It can be deduced in this example that gentleness, in the context of committee
chairmanship, is a bad thing, because unfailingly courteous to the point of gentle-
ness occurs as a concession countermanded by positive evaluation (fierce inde-
pendence and skill). However, the phrase to the point of provides the reader with a
short cut to arriving at this interpretation.
No similar explanation can be found for the first example. Examination of
a very wide context (the whole article in this case) can find no evidence that the
writer considers invigoration to be a negatively-evaluated quality. It is possible,
following Louw (1993), to suggest that the writer’s positive evaluation is in fact
insincere, and that to the point of invigoration is a slip that reveals this, but in the
absence of other evidence this must remain speculation. A more substantiated
argument is that this writer simply uses the phrase in an unusual way, without the
implications of the dominant use. This example indicates that we cannot simply
assume that every example of to the point of will indicate negative evaluation. On
the other hand, this use is so dominant that it can be used to argue that dramatic
to the point of tragedy has three indications of stance (dramatic, to the point of,
and tragedy) rather than two. The counter-example is a warning, however, that
attempts to quantify negative stance by counting phrases such as to the point of,
without examining each instance in its context, will not be wholly successful.
6.4 An increasingly accelerating pace
The phrase an increasingly accelerating pace is very different, in that a search of

the Bank of English reveals no examples of it, indicating that the phrase itself is
rare, and suggesting that there is no dominant usage to be called on. It can also
be pointed out that the phrase is an instance of tautology, with increasingly being
redundant in terms of the ideational meaning of the phrase. This suggests that
increasingly is being used to indicate stance rather than information. Looking at
examples in the Bank of English, however, shows that the adverb co-occurs with
words indicating positive evaluation (increasingly confident, increasingly attractive)
as well as with those indicating negative evaluation (increasingly difficult, increas-
ingly desperate). We therefore have to examine the concept of accelerating pace.
With this phrase, we cannot do something so simple as to look at the examples
in a corpus and count the good and the bad. Rather, we have to use the examples
to uncover the cultural presuppositions behind the phrase. For example, it is clear
that although a fast pace of life is usually bad:
They live life at an accelerated pace.” <hl> Missing; Come-on; Ge

He died as he lived, at breakneck pace. He was only 24, yet he was
and live our lives at such a fast pace,” recalled the woman execut
we are to living life at a frantic pace, it is very easy to live as
They live their lives at a frenetic pace that allows contact with th
said the family lived life at full pace, with the triplets and thei
that life is lived at such a hectic pace and people are so self-cent
population living at a hectic pace. The notion of belonging to
t 33, can no longer live at the hot pace set by today’s younger, str
has learnt to live with the jittery pace of development and capital
expect to live with his searing pace. In the same week that Cham
a slow pace of life is not necessarily good:
mortals who live life at a gentler pace. Open the door to new roman
to live with the STUS’s plodding pace. ‘It’s slower but it’s much
their lives organized at a snail’s pace, a situation the two rambun
setting lives at its own slow pace, most Ibizans preferring ju
and live life at their own yawning pace. Doctor Pain’ is acceptably
In many instances, in fact, the pace of life is not described as fast or slow as much
as appropriate: to humanity (at a human pace), to a period of life (at a child’s pace),
or to an individual (at your own pace):
44 Susan Hunston
nimals is living at the appropriate pace of its own biological clock

could be like if lived at a human pace. Just the odd contact every
It is at this pace, a normal living pace, that the elderly must func
enjoy living life at your own pace.” Initially, he preferred s
prefers to live and work at his own pace. This person doesn’t care f
But they live life at their own pace, a pace they’ve been used t
learn to live life at your own pace, and occasionally wear fals
There’s time to live at a child’s pace rather than being ruled by
An interpretation of this evidence is that what is highly valued is a pace of life that
is under the control of the person living it, whether that be fast and exciting or
slow and manageable. If we now look at accelerating pace (not necessarily pace of
life) we find mixed evidence – about half the examples indicate something bad:
(30) …have seen events move towards war at an accelerating pace…

(31) The administration sought a means to control the accelerating pace of Soviet
arms acquisition
and half indicate something good or neutral:
(32) He was quickly swept up in the studio’s accelerating pace

(33) …the accelerating pace of reform
In all the examples, though, the accelerating pace is always out of control. In none
of the examples – good or bad – is there an agent controlling the pace. In the only
example with an agent – The administration sought a means to control the acceler-
ating pace of Soviet arms acquisition – the agency is unsuccessful. In other words,
if we put together the results of these two studies, an accelerating pace of life is bad,
not because it is fast, but because a ‘good’ pace of life is one that is controlled, and
an accelerating pace of anything is not controlled.
7. A rhetoric of stance
The above discussion of a number of instances from example 23 confirms the view
that evaluative meanings are cumulative and occur across phrases in texts. It sug-
gests that the evaluative character of a text – the amount and the type of evaluative
stance that it expresses – depends not only on the presence of evaluative lexical
items but also on the use of phrases that can be shown to resound with intertextu-
al meaning. In other words, the opinionated style of example (23) does not depend
on its words, but on its phraseology. An example that has been examined in detail
above is the phrase to the point of. Other examples can be found, even in such a
short paragraph (Hunston 2004). One candidate is the phrase a very real concern
about. A corpus search for very real NOUN about shows that the nouns that occur
in this sequence include: concern, concerns, dismay, fears, misgivings, and reserva-
tions. This again indicates a degree of redundancy, in that the frame very real …
about restricts the choice of noun to the extent that its meaning is almost wholly
predictable (see also Tognini-Bonelli 2001: 118 on the meanings of real). Another
example is the sequence If…so be it. Sample concordance lines are shown below.
In each case, the if clause indicates an unfortunate, but unavoidable, occurrence.
The function of the whole is to indicate that an unfortunate side-effect of a rec-
ommended action is considered acceptable. The recurrence of the function of the
sequence in turn means that the stance it represents can be taken as given.
her turn. If she dies in the meantime, so be it. A JONES, Blackford

If we have to battle this out in court so be it. But, believe me, th
If it upsets their preparations then so be it. But Hearts have bee
If that means somebody gets booked, so be it. We must be competit
disease. If that means taking drugs, so be it. Moreover, there hav
if that means looking a bit cartoon-y so be it. It’s a parody of re
If it means going into the red then so be it. We can get back
 If the fans want to blame me then so be it,” he says. Of course
and if we have to go to court, then so be it.” A spokesman for Gu
life,” adding: “If that means prison, so be it.” Mr Leeson said he
for example. If that means complexity, so be it. As a result of this
much the better. If force is needed, so be it. In this context, Mr
them. If it lacks easy saleability, so be it -- the most interest
If we have to do it by being hip, so be it.” The agency has
and if we are soundly beaten, then so be it,” the manager said.
In short, example 23 contains several phrases that can be shown to recur in the
context of evaluative stance, even though they do not evaluate themselves. In
other words, what makes this paragraph stance-heavy is the presence of a very
real…about, if…so be it, to the point of, and the redundancy they represent, rather
than the words concern, tragedy, and so on themselves.
The investigation of this short text extract leads to a hypothesis: what dis-
tinguishes subjective (or stance-heavy) from objective (or stance-light) texts is not
the quantity of explicitly evaluative lexical items in each, but the embedding or
otherwise of those items in phraseologies, which frequently co-occur with stance.
Those phraseologies can be identified intuitively, but intuition in this regard is un-
reliable, and an examination of many instances of the target phrase is required to
corroborate the perception of its role in the text under investigation. A corpus can
provide just such a set of instances. This argument also implies an alternative view
46 Susan Hunston
of the role of quantitative information in studies of stance. Whilst the counting of

stance items, and the comparison between various corpora in this regard, remains
a viable and useful activity, it is clear that of necessity it provides a partial view of
stance. However, the quantitative, in the sense of giving information about ‘what
is usually said’ or ‘how a phrase is typically used,’ has an important role to play in
the qualitative activity of identifying stance elements.
8. Conclusion: Quantitative and qualitative corpus research
In this paper, I have tried to show that corpora can be used effectively to quantify
markers of stance – though this work must be complemented by a more qualita-
tive approach – and raw figures should be treated as the starting point of investi-
gation rather than its end point. I have also suggested that qualitative work using
corpora can show typicality of use and in doing so can enable us to identify stance
markers – particularly markers of evaluative stance – that were previously un-
known. These markers are typically phrases rather than individual words.
I have also tried to suggest that whereas computers are essentially used to
count items, in the case of corpora, that counting can be used in the service of
qualitative research as well as quantitative. Corpora present us with evidence for
intertextuality and lead us to conclusions that the meaning of a single instance is
dependent on the meanings of many other single instances.
Notes
1. As demonstrated in other contributions to this volume, especially the chapters by Du Bois,

Keisanen, and Haddington, stance meaning is also distributed across larger interactional units.
In addition to phraseologies, a broader qualitative analysis of stance likewise incorporates
turns-at-talk, interactional sequences, and intertextuality.
2. Both these corpora are components of the Bank of English corpus, jointly owned by the
University of Birmingham and Collins publishers. Other examples used in this paper are from
the same corpus unless otherwise stated.
3. Channell notes that the phrase seems to be infrequent in American English – perhaps the
essential pessimism it expresses is a purely British phenomenon.
4. These concordance lines are taken from the Bank of English corpus.
5. I am grateful to Robert Englebretson for disagreeing with me here, and so forcing me to
examine the evidence more closely.
6. The use of tragedy to indicate a type of play is excluded here.
References
Becher, T. 1989. Academic Tribes and Territories: Intellectual Enquiry and the Culture of Disci-
plines. Milton Keynes: Open University Press.
Becher, T. and Trowler, P. 2001. Academic Tribes and Territories. 2nd ed. Milton Keynes: Open
University Press.
Biber, D. 1988. Variation in Speech and Writing. Cambridge: Cambridge University Press.
Biber, D., Johansson, S., Leech, G., Conrad, S. and Finegan, E. 1999. Longman Grammar of
Channell, J. 2000. “Corpus-based analysis of evaluative lexis.” In Evaluation in Text: Authorial
Stance and the Construction of Discourse, S. Hunston and G. Thompson (eds.), 38–55. Ox-
ford: Oxford University Press.
Charles, M. 2004. The Author’s Voice in Academic Writing with Reference to Theses in Politics and
Materials Science. Ph.D. dissertation, Department of English, University of Birmingham.
son (eds.), 57–73. Oxford: Oxford University Press.
Francis, G. 1994. “Labelling discourse: An aspect of nominal-group lexical cohesion.” In Ad-
vances in Written Text Analysis, M. Coulthard (ed.), 83–101. London: Routledge.
Francis, G., Hunston, S. and Manning, E. 1996. Collins Cobuild Grammar Patterns 1: Verbs.
London: HarperCollins.
Francis, G., Hunston, S. and Manning, E. 1998. Collins Cobuild Grammar Patterns 2: Nouns
and Adjectives. London: HarperCollins.
Granger, S. 1998. Learner English on Computer. London: Longman.
Groom, N. 2004. “The scholarly, the sound and the thin: Evaluative adjectives and disciplinary
values in academic book reviews.” Postgraduate seminar, University of Birmingham.
Hunston, S. 2004. “Counting the uncountable: Problems of identifying evaluation in a text and
in a corpus.” In Corpora and Discourse, A. Partington, J. Morley and L. Haarman (eds.),
157–188. Bern: Peter Lang.
Hunston, S. and Francis, G. 1999. Pattern Grammar: A Corpus-Driven Approach to the Lexical
Grammar of English. Amsterdam: John Benjamins.
Hunston, S. and Thompson, G. (eds.). 2000. Evaluation in Text: Authorial Stance and the Con-
struction of Discourse. Oxford: Oxford University Press.
Lindquist, H. and Mair, C. (eds.). 2004. Corpus Approaches to Grammaticalization in English.
Amsterdam: John Benjamins.
Louw, B. 1993. “Irony in the text or insincerity in the writer? – The diagnostic potential of se-
mantic prosodies.” In Text and Technology: In Honour of John Sinclair, M. Baker, G. Francis
and E. Tognini Bonelli (eds.), 157–176. Amsterdam: John Benjamins.
Maley, A. 1999. “Surviving the 20th century.” English Teaching Professional 10: 3–7.
Matthiessen, C. 2006. “Systemic functional profiles of system and text: Investigations based on
texts, text archives, and corpora.” In System and Corpus: Exploring Connections, G. Thomp-
son and S. Hunston (eds.), 103–142. London: Equinox.
Semino, E. and Short, M. 2004. Corpus Stylistics: Speech, Writing and Thought Presentation in a
Corpus of English Writing. London: Routledge.
Sinclair, J. 1991. Corpus Concordance Collocation. Oxford: Oxford University Press.
Sinclair, J. 2004. Trust the Text: Language, Corpus and Discourse. London: Routledge.
48 Susan Hunston
Sinclair, J. and Coulthard, M. 1975. Towards an Analysis of Discourse: The English Used by Teach-
ers and Pupils. Oxford: Oxford University Press.
Teubert, W. 2004. “Language and corpus linguistics.” In Lexicology and Corpus Linguistics,
M.A.K. Halliday, W. Teubert, C. Yallop and A. Čermáková (eds.), 73–112. London: Con-
tinuum.
Tognini-Bonelli, E. 2001. Corpus Linguistics at Work. Amsterdam: John Benjamins.
Linking identity and dialect
through stancetaking
Barbara Johnstone
Carnegie Mellon University
1. Introduction1
Geographic mobility and new patterns of social interaction associated with the
globalizing new economy have resulted at the same time in dialect leveling (Mil-
roy 2002; Trudgill 1986) and, in some places, increased popular attention to re-
gional variation (Beal 1999; Dubois and Horvath 2002). One such place is the
Pittsburgh, Pennsylvania (U.S.) area, where talk about local identity very often
includes talk about the local dialect (Johnstone 2000a; Johnstone and Andrus
2005; Johnstone and Baumgardt 2004; Johnstone et al. 2002). When “authentic”
Pittsburghers or Pittsburgh activities are described or parodied, local speech is
almost invariably mentioned or performed. Souvenir vendors offer t-shirts and
sweat shirts, refrigerator magnets, shot glasses, and coffee mugs decorated with
lexical items thought to be local and other words spelled in such a way as to sug-
gest local pronunciations. A dictionary of Pittsburghese, as the variety is locally
known, has been continuously in print since 1982, and there is a copy in most
middle-class homes; and Pittsburghers and others go online to contribute to lists
of local expressions and discuss what the dialect means for the community. One
way to attribute a quintessentially local identity to a person is to label him or her
a yinzer, a word derived from the local variant of the second-person-plural pro-
noun, yinz. While much of this discourse about local speech arises in the context
of nostalgia for the city’s working-class industrial past, discourse about Pittsbur-
ghese also enters into youthful identity work. Pittsburghers in their 20s and 30s,
whether or not they make routine use of stereotypically local phonological or
lexical variants, sometimes refer to themselves as yinzers, and they perform local
identity by means of playful performances of local-sounding forms, both in face-
to-face interaction and in more public fora. In a playful nod to the New Yorker,
50 Barbara Johnstone
a new Pittsburgh literary magazine was named the New Yinzer, and in 2003 a
student conceptual artist produced (removable) stickers with brief definitions of
local terms and affixed them to bus-stop shelters and mailboxes in a project she
referred to as “guerilla linguistics.”
Heightened dialect awareness, like that seen in Pittsburgh, arises through dis-
cursive practices that call attention to and normativize regional forms, at the same
time refiguring their role in presentations and representations of local identity
(Johnstone and Andrus 2005). This paper explores the role of stancetaking in one
instantiation of one such practice: a conversation about local speech between a
sociolinguistic fieldworker and two lifelong residents of the Pittsburgh area. In
the conversation, the Pittsburghers deploy several strategies for epistemic stan-
cetaking: for making the implicit or explicit claim, that is, that they know enough,
from the appropriate sources, to describe and evaluate the local dialect. Some
such epistemic stancetaking moves invoke external sources of authority such as
published lists of local words and examples provided by other people represented
as authentic dialect speakers, while others invoke the speaker’s own competence
in the dialect. These include performances of local dialect forms and other allu-
sions to local identity. This strategy proves particularly effective in the interaction
in question: the speaker who represents herself as a competent speaker of the
dialect and who can perform local dialect forms gets the floor more often and
is oriented to as an expert on Pittsburgh speech. As a result, the other speaker,
who at first distances herself from speakers of the local dialect, eventually revises
her identity claim, recasting herself as a local dialect speaker who knows local
forms directly rather than by hearsay. In other words, she links her identity as a
Pittsburgher with competence in the local dialect because it is rhetorically use-
ful in this interaction to do so. The micro-rhetorical interactional exigency that
requires epistemic stancetaking drives the identity claim. Stancetaking is thus one
of the mechanisms through which dialect and local identity become linked in dis-
course. Repeated engagement in metalinguistic talk in which claiming the social
identity of a competent dialect speaker is useful for epistemic stancetaking serves
to strengthen and stabilize the idea that being a Pittsburgher means being able to
speak the local dialect.
Variationist sociolinguists in the Labovian tradition are coming to see “iden-
tity” as a useful explanatory dimension in accounting for some patterns of lin-
guistic variability and their role in language change. Analysts of discourse in in-
teraction in the tradition of Gumperz (1982) and Ochs (1992) find “stance” a
useful explanatory category in accounting for how particular linguistic choices
in interaction accomplish particular social and rhetorical actions. Sociolinguists
drawing on both traditions are beginning to explore how stancetaking can be
accomplished through phonological, morphological, and lexical choices, and
how sets of such choices can accrete into stances that index culturally meaningful
styles or identities (Eckert 2000; Kiesling 2005). This paper continues the work
of exploring how connections between linguistic forms and social identities are
shaped by interactional needs for stancetaking. With Eckert and Kiesling, I show
that stancetaking and identity are intertwined, in this case because adducing and
performing the social identity of a competent dialect speaker is a powerful re-
source for epistemic stancetaking, so much so that the need for epistemic stan-
cetaking can actually drive dialect-identity claims.
I begin by reviewing how the terms stance and identity have been used in
recent work by interactional sociolinguists and conversation analysts that aims
to account for what goes on in particular interactions, and how variationists have
adduced these concepts in accounting for patterns of variation across popula-
tions. I then show how dialect forms and regional identity can become linked
through stancetaking, by virtue of the fact that performances or evocations of
dialect competence can function as stancetaking moves. The text I use to illus-
trate this is an extended transcribed extract from a sociolinguistic interview. The
interview was conducted in the course of a project that brings together analyses
of regional patterns of phonological, morpho-lexical, and syntactic variation with
analyses of the real-time, interactive design of talk in particular rhetorical situa-
tions, by particular individuals. I describe how dialect identity – one’s positioning
as a user or nonuser of the local dialect – emerges as a rhetorical resource for and
through stancetaking in a conversation meant to probe Pittsburghers’ ideas about
the local dialect and explore how these ideas arise.
2. Identity and stance
Almost all work about social interaction that adduces the idea of stance as an
explanatory tool includes under stancetaking the moment-by-moment choices
speakers make that index their relationship to what they say (e.g., whether they
are sure or unsure about it, happy or sad about it, surprised or not). Building on
early work on epistemic and attitudinal stance (Biber and Finegan 1989, 1994;
Conrad and Biber 2000), Hunston and Thompson (2000) operationalize stance
as evaluation. Others use “stance” to talk about the marking and claiming of in-
terpersonal relations in talk as well. For Ochs (1992), particular linguistic forms
directly index evidential stances such as certainty, interpersonal stances such as
friendliness or intensity, or social actions such as apologizing. Particular stances
or social actions can then get linked indirectly to social identities such as gen-
der categories (so that, for example, in a particular sociocultural milieu, a stance
such as deference might become indexically linked with femaleness). For Du Bois
(this volume: 163), “Stance is a public act by a social actor, achieved dialogically
through overt communicative means, of simultaneously evaluating objects, posi-
tioning subjects (self and others), and aligning with other subjects, with respect to
any salient dimension of the sociocultural field”; the “sociocultural field” consists
of two social actors and an object to which both are oriented. Alignment or dis-
alignment with another social actor can be accomplished through membership
categorization moves; thus, claims to social identity for oneself and ascriptions of
identity to others fall under the rubric of stancetaking.
Like Goffman’s (1959) “presentations of self,” “identities,” in the interactional-
sociolinguistic tradition, are social categories to which speakers orient as they
become relevant in the interaction at hand. Some identities are “macrolevel de-
mographic categories,” while others are more situation-specific roles or “ethno-
graphically emergent cultural positions” (Bucholtz and Hall 2005: 585). Research
on social interaction that adduces identity as an explanatory tool almost always
includes under “identities” culturally circulating, frequently adduced ways of cat-
egorizing groups of people that are often oriented to as being relevant outside of
and prior to the interaction as well as inside it. “Identities” thus include ethnic,
class, and gender categories, and categorizations in terms of attributes such as
deviance vs. normalcy, tastes, and activities, as well as “discourse identities” such
as speaker or audience member. (Perhaps because of a misplaced fear of being
seen as locating social agency in the individual human, this body of literature
pays little attention to the way in which identities can be associated with indi-
vidually-embodied speakers (Johnstone 1996, 2000b).) But “identity work,” as it
is described in some of this literature, includes interactional moves that could also
be described as stancetaking. As Bucholtz and Hall point out, “identities may be
linguistically indexed through…stances” (2005: 585), and a repeated stancetaking
move or pattern of moves may emerge as an identity.
Interactionists in the conversation analysis tradition take a similar approach.
For them, identity arises in interaction: “for a person to ‘have an identity’…is to be
cast into a category with associated characteristics or features” (Antaki and Widdi-
combe 1998: 3). “Casting into a category” can be accomplished through stancetak-
ing. Conversation analysts stress the need to treat identity as interactional achieve-
ment with consequences for the structure of the talk, even if social actors think
identities pre-exist interactions and sometimes predict how they will play out.
3. Identity and stance in variationist sociolinguistics
Labovian-variationist sociolinguistics (Labov 1994, 2001) is primarily focused on

uncovering the mechanisms of language change. Variability is an inevitable con-
comitant of change, so in order to understand how language change is likely to

proceed, we need to understand how variability arises and what predicts its out-
come. The classic Labovian account uses the correlational methods of quantita-
tive sociology to model how “social facts” about speakers, such as socio-economic
status and sex, account for patterns of variation in groups. In the research settings
in which most early work was done, upper-class and female speakers consistently
tended to use more of certain variants than lower-class and male speakers did,
presumably because the variants in question were more “prestigious.” Variation
within an individual’s speech was linked with self-consciousness, which gives rise
in many cases to more careful speech, aimed at what speakers think of as more
prestigious variants.
From the beginning, however, in Labov’s own work as well as in others’, more
particular questions were being asked about why these kinds of patterns might
exist. Why, for example, might males use less “prestigious” forms than females?
What might make a working-class speaker adopt more local-sounding forms?
Although not framed in terms of stancetaking or identification, the answers that
were proposed suggested that phonological variability could be a resource for in-
dexing attitude and affiliation. In his groundbreaking study of Martha’s Vineyard,
Labov (1963) showed that the speakers who identified most closely with tradi-
tional local ways of living were most likely to use the more conservative, older, less
standard-sounding variants of certain phonological variables. This suggested that
the more local-sounding forms might in fact be part of the process through which
local identity was claimed. James and Lesley Milroy’s work in Belfast (J. Milroy
1992; L. Milroy 1987) explored the utility of social network theory for uncovering
and explaining patterns of variation. People with relatively many and overlap-
ping ties to their neighbors (such that they knew the same people in multiple
roles, and the people they knew also knew each other) were relatively likely to
talk in a more local-sounding, less standard-sounding way, because dense, multi-
plex social networks are effective enforcers of local norms. “Social class” was thus
reconceptualized in terms of social activity and from the perspective of speak-
ers’ everyday experience of language. Peter Trudgill’s (1972) equally influential
study of sex-correlated differences in Norwich, England suggested that men may
use more non-standard, local-sounding forms because such forms carry “covert
prestige” as signals of working-class solidarity. At the same time, R. B. LePage
and Andrée Tabouret-Keller (1985), concerned with describing the complicated
linguistic situation in the Caribbean, were suggesting that choices among variants
in a speaker’s repertoire could be thought of as “acts of identity,” and that speech
communities and linguistic varieties could be seen as ways of labeling the fact
that ways of identifying and, accordingly, of talking, sometimes become relatively
consistent.
Variationists doing quantitative, correlational work about speech commu-

nities have started to add “identity” to the lists of socio-demographic variables
they use to account for patterns of phonetic variability, finding that identity (vari-
ously operationalized) sometimes correlates more closely with variation than do
variables such as age, occupation, or place of residence. For example, Guy Bai-
ley and his colleagues (Bailey et al. 1993) show that, in Texas, linguistic changes
that diffuse from rural to urban settings typically involve the reassertion of tra-
ditional speech norms, arguably as badges of local identity. One such feature is
the monophthongization of /ay/ before voiceless obstruents, as in [ra:t] for right.
Speakers who do this are more likely than others to think highly of Texas as a
place to live. Hazen (2000) finds that North Carolina speakers with ties to institu-
tions and cultural characteristics from outside their county are more likely to shift
toward standard-sounding pronunciation in formal speech (perhaps because they
are more aware of the stigmatization of local speech), while people with more lo-
cal identities are linguistically more consistent. Rose (2003) has found that among
older people from the upper Midwest, the pronunciation of /θ/ and /ð/ as [t] and
[d] is not only correlated with rural residence but overtly commented on as a
symbol of a “country” identity. Schilling-Estes (1998) shows that in Ocracoke (an
Atlantic-coast U.S. island), a speaker self-consciously demonstrating the local
pronunciation of an expression he thinks of as particularly local performs the
role (Schilling-Estes does not use the term identity) of a stereotypical local person
with a more vernacular accent than his own. Schilling-Estes suggests that such
performances may be more patterned than variationists have assumed and may
reveal facts about dialect awareness than can affect the course of change.
Turning to stancetaking, interactional sociolinguists have explored how
choices among forms associated with different languages and dialects can signal
attitude and affiliation. For example, Ben Rampton (1995, 1999) has studied “lan-
guage crossing” or “styling the ‘other,’” exploring how shifts from a speaker’s “na-
tive” language or variety to and from one clearly associated with another group
can accomplish shifts in stance. Rampton (1995) describes how brief excursions
into Punjabi by London adolescents of Anglo and Afro-Caribbean as well as In-
dian and Pakistani descent can signal oppositional footing in interactions with
adults or key a joking stance in playground interaction. Stancetaking work can
also be accomplished through shifts into and from ways of speaking that are not
so clearly “other,” but rather part of a speaker’s native repertoire (Coupland 2001;
Rampton 2003). In Texas, for example, where sounding Southern is a resource
more or less “natively” available to many people, moves toward more Southern-
accented speech range, depending on the speaker and the situation, from fairly
automatic style shifts correlated with register to quite self-conscious rhetorical
moves, as when “Southern Belle”-sounding speech is used in footing-shifts meant

to manipulate men (Johnstone 1999a).
Penelope Eckert’s (1989, 2000) work in a Detroit high school suggests how
the variationist and interactionist approaches to stance, identity, and phonology
might be linked, suggesting how correlations like the ones described by Bailey,
Hazen, Rose, and others come to be. For Eckert, the choice between one variant
and another is part of the semiotic activity in which social identities are created,
not simply a reflection of already-existing differentiation. For example, adopting
certain variants of vowels is one way of adopting a stance toward and participat-
ing in local activity, just as is cruising in cars, dressing in a particular style, or do-
ing some things rather than others in school. Scott F. Kiesling (2005) shows that
a particular set of morpho-phonological features that co-occur in the English of
recent immigrants in Australia work together to project a face-saving epistemic/
interactional stance of “authoritative connection,” which, Kiesling claims, is par-
ticularly likely to be relevant for members of subordinate groups. Because they
work together as a stancetaking strategy, the features get used repeatedly together,
and a repeatable style (locally called wogspeak) emerges, linked with a repeatable
social identity, that of the recent immigrant. Eckert’s and Kiesling’s work shows
how linguistic variants can become indexically linked to social identities through
stancetaking. Particular linguistic features available in a speaker’s sociolinguistic
environment can be used for stancetaking, and sets of co-occurring stancetaking
features can come together as styles that index identities. The identities linked
with linguistic styles may be ethnic (as in Kiesling’s wogspeak, derived from a
derogatory term for non-whites); they may have to do with local social categories
(as with Eckert’s jocks and burnouts), or they may be linked with other sources
of identity. When the identity in question is regional, the style that indexes it is
often referred to as a dialect, by linguists and laypeople alike. In what follows, I
exemplify how stancetaking, identity, and linguistic variation are linked with ref-
erence to how ideas about what counts as a dialect are negotiated and circulated
in interaction.
4. Pittsburghese in conversation: Identity as a stance resource
Both linguists and non-linguist Pittsburghers associate a distinctive set of linguis-

tic features with southwestern Pennsylvania (Brown 1982; Hankey 1965, 1972;
Johnstone et al. 2002; Johnstone et al. 2004; McCarthy 2004; McElhinny 1999),
and Pittsburghers talk about this dialect often, in many contexts and media, fre-
quently in connection with talk about local identity (Johnstone 1999b; Johnstone,
Andrus, and Danielson 2006; Johnstone and Baumgardt 2004; Kiesling and Wis-
nosky 2003). The dialect appears to be receding: research on one feature, /aw/-
monophthongization, shows that it is now more common in the speech of older,
working-class speakers, the population that might be expected to use a receding
feature longest (Kiesling and Wisnosky 2003; McCarthy 2004). Local-sounding
talk, acquired early in life in face-to-face encounters, once identified people as
Pittsburghers only to occasional outsiders who noticed the dialect. Insiders, in
daily contact with people who sounded the same as they did, seldom noticed or
commented on local speech.
Beginning in the 1960s, however, out-migration caused by the collapse of
the steel industry and in-migration caused by the growth of the educational and
health-care sectors provided increased opportunities, in discursive contexts such
as nostalgic talk by ex-Pittsburghers and identity work by newcomers, for the
calling to attention of linguistic difference that creates heightened awareness of
regional varieties, and Pittsburghers started to use certain local speech features
to point to local identity in a more reflexive, stylized way (Johnstone, Andrus,
and Danielson 2006; Johnstone and Baumgardt 2004). The area thus lends itself
to a study of how dialect awareness arises in a variety of metalinguistic genres
of talk. As one phase in a larger study of Pittsburgh speech, co-workers and I
have conducted over 100 sociolinguistic interviews in four Pittsburgh-area neigh-
borhoods. The interview protocol elicits explicit talk about Pittsburghese, as it is
locally known, and, throughout the interviews and other research tasks, people
often break into spontaneous performances of the dialect. My analytical method
in this paper is discourse analysis, by which I mean close, systematic reading of a
small amount of text (Johnstone 2002).
In the interview extract that we will examine in detail, a woman in her 40s
and her 13-year-old daughter talk about Pittsburgh speech in an explicitly nor-
mative way. They offer examples of what Pittsburgh speech sounds like and argue
about which forms are authentically local and how local forms should sound. As
they do this, they use claims about and performances of their own speech as ways
of establishing the authority to describe the dialect, linking epistemic stance with
dialect identity.
To keep the transcript readable and avoid caricaturing Jen and Donna by
means of eye dialect, I have made notes about their accents in the right-hand col-
umn of the transcript rather than using the IPA transcription or respelling in the
text. BJ is the fieldworker. Simultaneous talk is linked with square brackets; equal
signs indicate “latched” talk. Italics mark loud or otherwise stressed words. Jen R.,
the mother, makes routine use of a number of phonological variables that make
her sound local. Her pronunciation of /aw/ is sometimes monophthongized, with
out realized as [a:t]. She merges and rounds the low back vowels (LBV), pronounc-
ing job as [jfb], for example. She fronts /u/ in words like move and sometimes vo-
calizes /l/. She reduces the diphthong /ay/ to a more monophthongal form when
it is followed by /l/ (which is vocalized), as when while is realized as [wa:u]. This
is another fairly common local-sounding variant. Her daughter Donna has a less
local accent; she does not monophthongize /aw/ but fronts /u/ and /o/, vocal-
izes /l/, and merges and rounds the low back vowel. As do most sociolinguistic
interviews in the Labovian tradition (Labov 1984), this one included modules on
topics meant to elicit a range of levels of self-consciousness. My summary of Jen
and Donna’s accents is based on the whole interview, as well as unrecorded talk
in other contexts. The topic of accent may well have made them self-conscious
about their speech, but in fact they sound very much the same in this segment as
elsewhere.
(1) FH01 (“Jen R”) and FH02 (“Donna R”) Pittsburghese

1 bj Um. So, have you ever heard of Pittsburghese?
2 jr Oh yes. I mean, [there’s]
3 bj [What- ]
4 jr that store over on the Southside in um, [Station Square that has diphthongal /aw/
the] Pittsburghese shirts in Southside
5 dr [((Breath intake)) Yeah]
6 bj Uh huh=
7 jr =In fact, I remember when my friend Karen moved out of state, fronted /u/
with- her husband’s job took them out of state and to many in moved;
other states. I remember sending her a couple Pittsburghese monophthongal
shirts for them. /aw/ in out
(twice); rounded
low back vowel
(LBV) in job;
“dropped g” in
sending
8 dr Hm
9 jr Umm. Yeah, I’ve heard of [Pittsburghese,] definitely. Yeah.=
10 bj [Mm hmmm.] =What do you think it
is? I mean is it-?
11 jr I think it’s the way we say words.
12 dr [((laughs))]
13 bj [Yeah?]
14 jr I think it’s how we say “downtown” and um= diphthongal /aw/s
in downtown
15 dr = “down[town”] diphthongal /aw/s
16 jr [“South]side” and, diphthongal /aw/
in Southside
17 dr “Y’all”
18 jr “wash” and “iron” and different words and the way Pittsburgh “standard,”
is. There’re not local
pronunciations of
wash, iron
19 dr “Yinz” [yHnz]
20 jr [Just the uh- like]
21 bj [“Yinz” is another one.] What, what other ones can you think [yHnz]
of?
22 dr Just “y’all” and “yinz.” That’s, that’s the most my friends always
are saying “y’all” to me.
23 bj [“Y’all?”]
24 dr [Drives] me crazy.
25 jr “Y’all?”
26 dr Yeah, they say “y’all” to me. They say [it’s a Pittsburghese]
27 bj [And that’s a Pittsburgh
thing?]
28 dr It- that’s what they tell me.
29 bj Huh!
30 jr “Younz” is more a Pittsburgh thing than [“y’all.”] [y~nz]
31 dr [yeah]
32 jr “Y’all”’s more like a Georgia, [Southern.]
33 dr [I was thinking] Southern, [but ]
34 jr [Yeah]
35 dr they still say “y’all” to me. And then. Yeah “yinz.” You hear LBV not rounded
“yinz” a lot. in lot
36 bj You do? In-
37 dr Yeah well like our neighbors like two doors down, I’m really fronted /u/ in two,
good friends with their son. He’s a year older than me. And like, do;
he says “yinz” constantly, ‘cause like both his parents say “yinz,” vocalized /l/ in
like “Yinz wanna do somethin’?” or [like] ((laughing)) you know older
so rounded LBV in
[I hear that.] constantly; fronted
/o/ in both, know
38 bj [Mm hmm]
39 jr [We don’t use that.]
40 dr Yeah I never said you [used it] but
41 jr [Yeah,]
42 bj You don’t use that. [Uh huh.]
43 jr [but] I’m just thinking, I know, um, like, your
dad and I don’t use that [too often.]
44 dr [No.] But I hear it a lot from [them LBV not rounded
when] I’m over there in lot; /o/ less
fronted than
previously in over
45 jr [Mm hmm]
46 bj [Mm hmm]
47 jr Yeah. It’s funny.
48 dr And you pick up on it. You start to say then, [once]

49 jr [Sure.]
50 dr you’re around people so often, you start- I started to say “yinz” diphthongal /aw/
to people. ((laughing)) And they’re looking at me like, “Okay.” in around
((skeptical, amused voice))
51 dr Mm hmm
52 bj So, would- you don’t- you wouldn’t say you use any of the,
Pittsburghese things?
53 dr Not really. No, I don’t think so.
54 bj Uh huh [(probably you-)]
55 jr [You s-] you do, but you don’t reali-. Oh, I know I do. I
[say “iron”] iron = [arn]
56 dr [I do?]
57 jr for “iron.” iron = [ay6rn]
58 bj Uh huh.
59 dr [“iron”] [arn]
60 jr [I] say “wash w- wash” for “wash”. I [mean] [w~š] [wfš] [wfš]
61 dr [“wash”] [wfš]
62 jr I don’t, I don’t pronounce my words as clearly as-,
63 dr “wash” [wfš]
64 jr or the accent’s [on]
65 bj [uh huh]
66 jr a different [part]
67 dr [“wash”] [wf]
68 bj Uh-huh
69 jr Yeah
70 dr “wash” [wfš]
71 jr “Southside” instead of “Southside.” monophthongal
/aw/ and
assimilated
/θ/ in first,
“performance,”
Southside;
exaggerated
diphthongal
/aw/ and /θ/ in
second, “citation”
Southside
72 dr “Southside” Monophthongal
/aw/
73 jr I say “Southside.” monophthongal

“Downtown” instead of “downtown.” /aw/
monophthongal
/aw/s in first,
“performance,”
downtown;
exaggerated
diphthongal
/aw/s in second,
“citation”
downtown
74 dr “downtown” diphthongal /aw/s
75 jr I know I, I know I kn- use a lot of Pittsburghese.
76 bj Mm hmmm.
77 dr I [probably do] and I don’t realize it fronted /u/ in do
78 jr [I know I do.] Well when I’ve been in [different] states, in
different cities,
79 bj [Mm hmm]
80 dr They’ll- they’ll say “You’re from Pittsburgh.”
81 jr Yeah, they’ll immediately [say,]
82 dr [When we were in] South Carolina, diphthongal /aw/
right? in South
83 jr “You’re from Pennsylvania=”, /l/ not vocalized
in Pennsylvania
84 dr “=Yeah. you’re- you’re [definite-]”
85 jr [“Are you-], are you from the Pittsburgh
area?”
86 dr Yeah. ((laughing))
87 bj Does that happen to you, too, or=?
88 dr =Yeah. I mean, I remember one time, we were in South Carolina diphthongal
visiting my, my uncle and my two cousins and my aunt. And we /aw/ in South;
went to (s- some) store, and we were talking about how like the vocalized /l/ in
South kind of moves slow, [you know?] uncle; fronted /u/
in two, moves;
diphthongal /aw/
in about, how;
monophthongal
/aw/ in South.
89 jr [Yeah, ] [God, it drives you
crazy.]
90 dr [And then she’s like] “Pennsylvania
she’s like, “You guys from Pennsylvania?” We’re like “Yeah.” And question”
she’s like “You guys wouldn’t happen to be from Pittsburgh, intonation;
right?” And we’re like, “We’re from Pittsburgh.” And she’s like, vocalized /l/ in
“Oh, okay. I can [tell by your accent.”] Pennsylvania
91 bj [How did she know?] How do you think she
knew?
92 jr She, she was from the Pittsburgh area. She didn’t grow up in didn’t = [dι’6n]
Pitt- [yeah. She said, she said she originally]
93 dr [Yea-, I thought she was] from Ohio.

94 jr Yeah, [but she she was,]
95 bj [(Eastern)] Ohio?
96 jr Right [but she]
97 bj [This area?]
98 jr she was in Pittsburgh and then they moved to Ohio, fronted /u/ in
moved
99 dr Yeah
100 jr and then from Ohio they moved to South Carolina. fronted /u/
in moved;
monophthongal
/aw/ in South
101 dr [She didn’t have the] accent either. She still had like a Pittsburgh
accent.
102 jr [She said that-] Mm hmm. Yeah. She didn’t [have] didn’t = [dι’6n]
103 bj [Mm hmm]
104 jr a Southern accent at all. She was, like a mom. [You know,] rounded LBV in
mom
105 bj [Mm hmm]
106 jr I would say she was in her early 30s at least.
107 dr Yeah.
108 jr [Uhh maybe 40s.]
109 dr [ I don’t really remember]
110 jr I don’t know. I- I’m a bad judge of age. But, she was from, eh or glide-reduced /ay/
grew up in Pittsburgh for a while, moved to Ohio, and then was in I; vocalized /l/
in South- then they relocated to South Carolina. in while; fronted
/u/ in moved;
monophthongal
/aw/ in South
111 dr [yeah]
112 jr [That’s] how she knew. ‘Cause she even said that Pittsburgh
accent, when you’re not a-round it, when you do hear it, you diphthongal /aw/
really pick up on it fast. in around
Epistemic stancetaking and dialect identity come into play here in a number of
ways. At the beginning of the extract, Jen claims the authority to speak on the
topic of Pittsburghese with reference to knowing about Pittsburghese shirts and
sending one to a friend who has moved away. The shirts Jen is referring to, pro-
duced largely for the tourist and local-nostalgia markets, feature words spelled in
ways that suggest their “Pittsburgh” pronunciation; on the back, there may be a
dictionary-like list of words and phrases thought to be local. Epistemic stancetak-
ing is independent of dialect identity here. Referring to Pittsburghese shirts is a
way of arguing from external authority, a resource that is potentially available
whether or not one is a speaker of the dialect. Jen supports her epistemic claim
(Oh yes [I’ve heard of Pittsburghese], line 2) with reference to indirect, mediat-
ed knowledge about the dialect – she has seen it on t-shirts. She maintains this
relatively detached epistemic stance for another turn (line 9), Yeah, I’ve heard of
Pittsburghese, definitely. I’ve heard of locates the epistemic source elsewhere, in
what other people say. Then, however, in response to my question What do you
think it is?, Jen switches to a different mode of evidence, taking up my invitation
to adopt an epistemic stance rooted in personal authority (I think) and aligning
herself with other competent speakers of the dialect (we), I think it’s the way we
say words.
Jen then begins to list some of these (lines 14–18), downtown, Southside, wash,
iron. While continuing to locate the source of knowledge in her own competence
(I think), she disaligns somewhat from dialect speakers and returns to a more dis-
tanced mode of epistemic stancetaking that does not rely on competent-speaker
dialect identity. The citation forms she produces are not the local pronunciations;
anyone who has read or heard about Pittsburghese could produce them, whether
or not they knew how they sounded when pronounced by someone with a local
accent. Jen pronounces /aw/ as a diphthong in downtown and Southside, using the
less local-sounding variant. She also pronounces wash and iron in the standard
ways, rather than in the local-sounding variants [w~š] or [worš] and [arn]. Nor
do these citation forms fully reflect what Pittsburghers usually imagine is local
about “the way we say [these] words.” Downtown is typically spelled “dahntahn”
on artifacts like t-shirts, the spelling suggesting that the monophthongization of
/aw/ is to be attended to, whereas Southside is often spelled “souside,” with a diph-
thongal /aw/ but a deleted or assimilated /θ/.
In lines 71 and 73, Jen returns to two of these words, contrasting what she
represents as their standard pronunciation with the way she claims to say them. In
citing the “standard” forms, she exaggerates the diphthongs in both words and the
/θ/ in Southside. In her performance of her own pronunciation, she overdoes what
popular local spellings suggest are the local pronunciations, monophthongizing
the /aw/ in both words rather than just in downtown. Here, Jen claims an authori-
tative stance in two ways. In citing examples of Pittsburghese in their standard
pronunciations, she is doing something that either a speaker or a non-speaker
of the dialect could presumably do, assuming he or she had access to lists of lo-
cal forms like those on t-shirts or folk dictionaries. In this activity, authoritative
stance is independent of dialect identity. But Jen also performs the local pronun-
ciations, an activity that indexes the dialect identity of a competent speaker. She
also points to this competent-speaker identity repeatedly in more explicit claims
to be an actual user of the dialect, We don’t use that…I’m just thinking…your dad
and I don’t use that too often (lines 39–43), I know I [use the Pittsburghese things]
(line 55), I don’t pronounce my words as clearly…(line 62), I know I, I know I kn-
use a lot of Pittsburghese (line 75). Note how in this final extract Jen starts to say
she knows Pittsburghese, which could signal second-hand access to the dialect,
but revises the claim to I use a lot of Pittsburghese, explicitly claiming to speak it.
To summarize, Jen makes epistemic stancetaking moves throughout the con-
versation. Some of these involve displaying familiarity with external sources of
authority such as Pittsburghese shirts. Some of these moves involve performances
of this knowledge, in the form of citations of local forms in a standard-sounding
way. Other stancetaking moves involve direct claims to competent-speaker dia-
lect identity. Sometimes, as we saw above, performances of competent-speaker
dialect identity are embedded in these claims, I say [arn] for [ay6rn] (lines 55–57)
and I say [w~š] for [wfš] (line 60).
Donna, the 13-year-old, tries to participate in all these activities. At first, her
epistemic stancetaking is marked by moves that distance her from dialect speak-
ers and locate the source of her knowledge about the dialect in others. Invited,
like her mother, to talk about what she thinks Pittsburghese is, she talks about
what other people say it is. In line 17, she suggests an addition to the list Jen is
building, y’all, then, after there is no uptake from her mother or me, another in
line 19, yinz. I acknowledge this contribution in line 21 and encourage her to offer
more. She repeats y’all and yinz in line 22, then explicitly adduces the source of
her epistemic authority on the topic of local speech, my friends always are saying
“y’all” to me and They say it’s a Pittsburghese [thing] (line 26). When she continues
to be met with skepticism, she makes the same stancetaking move again: It- that’s
what they tell me (line 28). These stancetaking moves are not linked to compe-
tent-speaker identity – Donna would have access to this source of knowledge
whether or not she claimed to be a speaker of the dialect herself – but rather to
external authority. Her mother then rebuts Donna’s externally-based claim with
a dialect-performance move: arguing that y’all is not really “a Pittsburgh thing,”
she pronounces yinz not in the stereotypical version represented on t-shirt lists,
which would be [yHnz], but in an older, more traditional-sounding way, [y~nz].
She then supplements this with a more distanced move referring to presumably
widespread knowledge that does not require dialect identity, Y’all’s more like a
Georgia, Southern [thing] (line 32). Donna continues to argue that y’all is local, but
continues to disalign herself from the local way of talking, contrasting I with they
and positioning herself as the recipient of local speech rather than its initiator, I
was thinking Southern, but they still say “y’all” to me (lines 33–35). But her mother’s
competent-speaker knowledge apparently trumps Donna’s external knowledge:
Donna retreats to a discussion of yinz, which everyone in the interaction agrees is
local, Yeah “yinz.” You hear “yinz” a lot (line 35). Using you hear rather than I hear,
she aligns herself, if not with dialect speakers, at least with a group larger than
herself. She then provides an extended illustration of her claim to hear yinz a lot,
which includes a dialect performance (“Yinz wanna do somethin’?” line 37). Note
that this is not the same sort of dialect performance as Jen’s have been: Donna is
imitating other people, not making a claim about her own dialect identity. A per-
formance like this displays local knowledge (she knows how local speech sounds)
but falls short of a claim to being a speaker of the dialect herself.
As the conversation proceeds, Donna begins to supplement epistemic moves
that appeal to external sources with her mother’s interactionally more successful
mode of stancetaking by evoking a competent-speaker dialect identity. In her first
claim to actually being a dialect speaker, Donna frames her competence as an
unintentional and uncharacteristic consequence of being around dialect speak-
ers, .. you pick up on it. You start to say then once you’re around people so often, you
start- I started to say “yinz” to people. ((laughing)) And they’re looking at me like,
“okay” ((skeptical, amused voice)) (lines 48–50). In answer to my direct question,
however, she then explicitly disaligns herself from other dialect speakers, Not re-
ally. No, I don’t think [I use any of the Pittsburghese things] (line 53). Her mother
steers her toward alignment with dialect speakers, You do, but you don’t real[ize
it] (line 55).
Donna appears to take the hint. She begins to reframe her dialect identity in
such a way that it becomes useful in epistemic stancetaking, the way Jen’s dia-
lect identity is. One revealing segment begins in line 60, where Jen, listing and
performing local forms, makes and tries to illustrate a claim about how she says
wash. In an apparent performance error, she almost confuses the “correct” form
with the “Pittsburghese” form, so that the second time she says the word it sounds
like the standard [wfš] but is apparently meant to be an improved performance
of what Jen represents as the local pronunciation, [w~š]. Donna, who has just
claimed that she does not use Pittsburghese things (line 53), then starts to repeat
the word over and over in lines 61, 63, 67, and 70, in a low voice, apparently try-
ing to imitate the local pronunciation so as to contrast it with her own. But since
Donna picks as her target Jen’s second performance, which was actually the more
standard-sounding variant, Donna seems to conclude that her own pronuncia-
tion is in fact the local one. So after “trying out” Southside and downtown in a
similar manner, she explicitly claims the identity of a dialect speaker in line 77,
echoing her mother’s earlier wording, I probably do [use Pittsburghese] and I don’t
realize it.
Shifting identity in this way means that Donna can now adopt the epistemic
stance of an actual dialect speaker, which her mother has been drawing on, and
she does this in co-narrating the story about the family’s having been identified as
Pittsburghers by their accents. This begins as an explicit claim, co-constructed by
Jen and Donna, to the identity of a recognizable dialect speaker:
78 jr [I know I do.] Well when I’ve been in [different] states, in different cities,
79 bj [Mmhmm]
80 dr They’ll- they’ll say “You’re from Pittsburgh.”
Jen then claims she is also recognized by her accent, Yeah, they’ll immediately say
“You’re from Pennsylvania” (lines 81–83). The ensuing narrative, which involves
densely overlapped joint production by Jen and Donna, supports their now mutu-
al claim to competent-speaker dialect identity. It culminates with Donna’s voicing
of her family (we) and a woman they met in the South (she): And we’re like, “We’re
from Pittsburgh.” And she’s like, “Oh, okay. I can tell by your accent” (line 90).
5. Discussion
I began this paper by noting that, in Pittsburgh, local identity and local dialect are
often linked, and by asking how such links are forged. How does being a Pittsbur-
gher get associated in so much popular discourse with speaking “Pittsburghese”?
I have explored one way this can happen: if people are talking about local speech,
then it can be interactionally useful to claim the identity of a local dialect speaker,
because doing so provides one with resources for epistemic stancetaking.
This conversation illustrates how both dialect identity and epistemic stan-
cetaking arise in interaction, in response to particular prompts (such as my Have
you ever heard of Pittsburghese? What do you think it is?) and more general inter-
actional exigencies such as wanting to get the floor. The two are intertwined in a
particularly visible way here. Since both the interactional genre (the interview) and
the particular topic called for knowledge claims and displays of the authority to
make such claims, epistemic stancetaking was an interactional requirement. Since
the topic was local speech, claims about and performances of competent-speaker
dialect identity were a particularly useful way to make epistemic stancetaking
moves. To get a sense of how stancetaking works in the conversation, I explored
both explicit claims to epistemic authority and indirect claims to such authority
via citations of local words and sounds. To see how and when dialect identity
becomes relevant in the conversation, I described explicit moves that characterize
the participants as speakers of the dialect (I talk that way) or not (I don’t really use
Pittsburghese things), and indirect claims to local-speaker identity through per-
formances of the local accent. As we have seen, epistemic stancetaking moves in
this conversation are often scaffolded on allusions to and performances of dialect
identity. The micro-rhetorical exigencies that require stancetaking can be seen to
drive identification moves, as when Donna recasts her identity in order to assume
a more authoritative epistemic stance.
That the topic of local speech came up in this case is not surprising: I brought
it up, as a module in a sociolinguistic interview. But the topic comes up nowadays
in many ways. As I have shown in this paper, once the topic arises, dialect can
become linked with local identity via the interactional usefulness of representing
oneself as a speaker of the dialect. It should be stressed, however, that not every-
one has the same kind of access to this resource. There is an important sense in
which Jen has a stronger local accent than Donna does, in part for linguistic and
cognitive reasons that are not related to identification or stancetaking. Claim-
ing to be a speaker of the local dialect is not the same as being one in the sense
linguists usually have in mind; it does not require anything more than knowing a
few local-sounding words. A person like Jen, who can perform local pronuncia-
tions, may have interactional resources in certain contexts that Donna, who can
only say she speaks the dialect, lacks. Discursive activities like the one examined
here give the upper hand to more competent speakers of the local dialect. This
means that competent dialect speakers like Jen have the stancetaking advantage
in this conversation and ones like it. It would be an oversimplification, however,
to suppose that its usefulness in discursive activities like this will automatically
contribute to the maintenance of the dialect in the face of powerful homogenizing
pressures. That is a hypothesis that remains to be tested.
Notes
1. Work on this project was partially supported by the National Science Foundation, Award
BCS-0417657. Jennifer Andrus helped with the transcription, and other co-workers on the
Pittsburgh Speech and Society project, in particular Scott F. Kiesling, have helped by discussing
earlier iterations of this paper with me. Comments on a different version of this chapter by fel-
low participants in the 10th Biennial Rice University Linguistics Symposium, “Stancetaking in
Discourse,” as well as comments on this version by Robert Englebretson, have been most helpful.
I am especially grateful to “Jen,” “Donna,” and other members of their family for their generosity
in talking to me and providing other kinds of invaluable assistance with the project.
References
Antaki, C. and Widdicombe, S. (eds.). 1998. Identities in Talk. London: Sage.

Bailey, G., Wikle, T., Tillery, J. and Sand, L. 1993. “Some patterns of linguistic diffusion.” Lan-
guage Variation and Change 3: 359–390.
Beal, J.C. 1999. “‘Geordie nation’: Language and regional identity in the north-east of England.”
Paper presented at Methods X. St. John, Newfoundland, Canada, August 1–6.
of evidentiality and affect.” Text 9: 93–124.
Biber, D. and Finegan, E. (eds.). 1994. Dimensions of Register Variation. New York: Oxford
University Press.
Brown, C. 1982. A Search for Sound Change: A Look at The Lowering of Tense Vowels before
Liquids in the Pittsburgh Area. MA thesis, University of Pittsburgh.
Discourse Studies 7: 585–614.
Conrad, S. and Biber, D. 2000. “Adverbial marking of stance in speech and writing.” In Eval-
uation in Text: Authorial Stance and The Construction of Discourse, S. Hunston and G.
Thompson (eds.), 57–73. New York: Oxford University Press.
Coupland, N. 2001. “Dialect stylization in radio talk.” Language in Society 30: 345–375.
Dubois, S. and Horvath, B.M. 2002. “Sounding Cajun: The rhetorical use of dialect in speech
and writing.” American Speech 77: 264–287.
Eckert, P. 1989. Jocks and Burnouts: Social Categories and Identity in High School. New York and
London: Teachers College Press.
Eckert, P. 2000. Linguistic Variation as Social Practice. Oxford: Blackwell.
Goffman, E. 1959. The Presentation of Self in Everyday Life. Garden City, NY: Doubleday An-
chor Books.
Gumperz, J.J. 1982. Discourse Strategies. Cambridge: Cambridge University Press.
Hankey, C.T. 1965. “Miscellany: ‘Tiger,’ ‘tagger,’ and [ai] in western Pennsylvania; Diphthongal
variants of [e] and [æ] in western Pennsylvania.” American Speech 40: 226–229.
Hankey, C.T. 1972. “Notes on west Penn-Ohio phonology.” In Studies in Linguistics in Honor
of Raven I. McDavid, Jr., L.M. Davis (ed.), 49–61. University, AL: University of Alabama
Press.
Hazen, K. 2000. Identity and Ethnicity in the Rural South: A Sociolinguistic View Through Past
and Present Be. [Publications of the American Dialect Society 83] Durham, NC: Duke
University Press.
Hunston, S. and Thompson, G. (eds.). 2000. Evaluation in Text: Authorial Stance and the Con-
struction of Discourse. New York: Oxford University Press.
Johnstone, B. 1996. The Linguistic Individual: Self-expression in Language and Linguistics. New
Johnstone, B. 1999a. “Uses of Southern speech by contemporary Texas women.” Journal of So-
ciolinguistics 3: 505–522. [Theme issue on Styling the ‘other,’ Ben Rampton (ed.)]
Johnstone, B. 1999b. “How to speak like a Pittsburgher: Representations of speech and the
study of linguistic variation.” Plenary address presented at the Sociedad Argentina de
Lingüística, Mar Del Plata, Argentina.
Johnstone, B. 2000a. “Representing American speech.” American Speech 75 (Diamond Anni-
versary Essays): 390–392.
Johnstone, B. 2000b. “The individual voice in language.” Annual Review of Anthropology 29:
405–424.
Johnstone, B. 2002. Discourse Analysis. Malden, MA: Blackwell..
Johnstone, B., Andrus, J., Baumgardt, D., Schardt, A.M. and Kiesling, S.F. 2004. “Whose social
meaning? Pittsburgh monophthongal /aw/ in perception and production.” Paper present-
ed at New Ways of Analyzing Variation 33. Ann Arbor, MI, September 30–October 3.
Johnstone, B., Andrus, J., and Danielson, A.E. 2006. “Mobility, indexicality, and the enregister-
ment of ‘Pittsburghese’.” Journal of English Linguistics 34: 77–104.
Johnstone, B. and Baumgardt, D. 2004. “‘Pittsburghese’ online: Vernacular norming in conver-
sation.” American Speech 79: 115–145.
Johnstone, B., Bhasin, N. and Wittkofski, D. 2002. “‘Dahntahn Pittsburgh’: Monophthongal

/aw/ and representations of localness in southwestern Pennsylvania.” American Speech 77:
148–166.
Kiesling, S.F. 2005. “Variation, stance, and style: Word-final -er, high rising tone, and ethnicity
in Australian English.” English World Wide 26: 1–44.
Kiesling, S.F. and Wisnosky, M. 2003. “Competing norms, heritage prestige, and the decline
of /aw/-monophthongization in Pittsburgh.” Poster presented at New Ways of Analyzing
Variation 32, Philadelphia, PA, October 9–12.
Labov, W. 1963. “The social motivation of a sound change.” Word 19: 237–309.
Labov, W. 1984. “Field methods of the project on linguistic change and variation.” In Language
in Use: Readings in Sociolinguistics. J. Baugh and J. Sherzer (eds.), 28–66. Englewood Cliffs,
NJ: Prentice-Hall.
Labov, W. 1994. Principles of Linguistic Change: Internal Factors. New York: Blackwell.
Labov, W. 2001. Principles of Linguistic Change: Social Factors. Malden, MA: Blackwell.
LePage, R.B. and Tabouret-Keller, A. 1985. Acts of Identity: Creole-Based Approaches to Lan-
guage and Ethnicity. Cambridge: Cambridge University Press.
McCarthy, C. 2004. “Language change in Pittsburgh: The decline of /aw/-monophthongization
and the Canadian shift.” Poster presented at New Ways of Analyzing Variation 33, Ann
Arbor, MI, September 30–October 3.
McElhinny, B. 1999. “More on the third dialect of English: Linguistic constraints on the use of
three phonological variables in Pittsburgh.” Language Variation and Change 11: 171–195.
Milroy, J. 1992. Linguistic Variation and Change. Oxford: Blackwell.
Milroy, L. 1987. Language and Social Networks, 2nd ed. Oxford: Basil Blackwell.
Milroy, L. (ed.). 2002. “Investigating change and variation through dialect contact.” [Theme is-
sue]. Journal of Sociolinguistics 6: 3–130.
Ochs, E. 1992. “Indexing gender.” In Rethinking Context: Language as an Interactive Phenom-
enon, A. Duranti and C. Goodwin (eds.), 335–358. New York: Cambridge University
Press.
Rampton, B. 1995. Crossing: Language and Ethnicity among Adolescents. London: Longman.
Rampton, B. (ed.). 1999. “Styling the ‘other.’” [Theme issue]. Journal of Sociolinguistics 3: 421–
556.
Rampton, B. 2003. “Hegemony, social class and stylisation.” Pragmatics 13: 49–83.
Rose, M. 2003. “‘On de farm’: Sociolinguistic meaning in town and country.” Paper presented at
New Ways of Analyzing Variation 32, Philadelphia, PA, October 9–12.
Schilling-Estes, N. 1998. “Investigating ‘self-conscious’ speech: The performance register in Oc-
racoke English.” Language in Society 27: 53–83.
Trudgill, P. 1972. “Sex and covert prestige: Linguistic change in the urban dialect of Norwich.”
Language in Society 1: 179–195.
Trudgill, P. 1986. Dialects in Contact. Oxford: Basil Blackwell.
Grammatical resources for social purposes
Some aspects of stancetaking in colloquial
Indonesian conversation
Robert Englebretson
Rice University
1. Introduction1
This paper is predicated on two points of departure: First, that stancetaking is

a pervasive activity which speakers engage in through the use of language; sec-
ond, that grammar is motivated and shaped by language use. Assuming these two
propositions, then, we should expect grammar itself to be directly implicated in
stancetaking. In other words, if stancetaking is a frequent activity of language use,
and if frequent activities of language use themselves play a part in shaping lan-
guage form, then it follows that stancetaking must therefore also be understood
as having a role in shaping language form, and the grammatical resources of a
language should likewise reflect principles of stancetaking. This paper examines
three such resources in colloquial Indonesian, based on a corpus of spontaneous
conversational data. I demonstrate how Indonesian speakers use first-person-sin-
gular referring expressions, the -nya clitic, and verbal diathesis (voice) respec-
tively to manage and index three facets of stancetaking in everyday conversational
interaction: identity, epistemicity, and positioning. I further argue that traditional
descriptions and accounts have overlooked crucial aspects of the meaning and
function of these three grammatical resources. Adopting a view of grammar as
rooted in social interaction (in this case, specifically, the social and interactional
processes of stancetaking) provides a rich set of previously undocumented obser-
vations about their meaning and use.
Regarding the first point of departure mentioned above, the ubiquity of
stancetaking in its various forms has been noted for several decades. As Stubbs
points out:
whenever speakers (or writers) say anything, they encode their point of view to-
wards it …The expression of such speakers’ attitudes is pervasive in all uses of
language. All sentences encode such a point of view, … and the description of the
markers of such points of view and their meanings should therefore be a central
topic for linguistics. (1986: 1)
In other words, to roughly paraphrase, every utterance enacts a stance, and aware-
ness of this should inform the linguist’s work at all levels.2 In the past 20 years,
since Stubbs’s proposal, numerous linguists have indeed taken up this call, es-
pecially those working in corpus-based, systemic-functional, sociocultural, and
interactional approaches to language. Yet, in general, most of this work has been
focused on English, and a recognition of the centrality of stancetaking has by and
large not found its way into field linguistics, nor into the writing of descriptive
grammars of other languages. For example, while there are numerous grammati-
cal descriptions available for both formal and colloquial varieties of Indonesian,
none of these sources has addressed stancetaking per se, nor have they addressed
the lexico-grammatical means which speakers use to accomplish it. Rather, In-
donesian descriptive and pedagogical grammars tend to reflect the general bias
in descriptive/typological/field linguistics which conceives of grammar as (pri-
marily, if not solely) a cognitive object and referential system. The present paper
hopes to expand this view by offering some observations on the social/interac-
tional nature of Indonesian grammar, and seeks to initiate a discussion of how
Indonesian speakers take stances. For the three aspects of Indonesian grammar
that I discuss in this paper, I show that, in addition to fulfilling their traditional,
cognitively-based referential functions of expressing and managing information,
they have specific interactional functions too, which contribute directly to the
social worlds speakers are constructing through stancetaking. There are undoubt-
edly more than just these three resources for stancetaking in colloquial Indone-
sian. For example, Wouk (1998, 2001) analyzes two clause-final particles used in
expressing solidarity in social interaction. The present paper obviously cannot
attempt a comprehensive and thorough analysis of the entirety of stancetaking in
the language. However, as a first step toward this goal, I offer these observations
and analyses in hopes that they will spark future research by other Indonesian
scholars interested in investigating stancetaking more thoroughly.
The second point of departure for this paper, the functionalist principle that
grammar is motivated and shaped by language use, has likewise informed the
work of numerous researchers over the past several decades. Language form has
been argued to be determined largely by the cognitive and biological makeup
of human beings, and the communicative contexts of language use (cf. Cum-
ming and Ono 1997; Ford et al. 2002; Givón 1979; Langacker 1999; Ochs et al.
Stancetaking in colloquial Indonesian conversation 71
1996; Ono and Thompson 1995; Tomasello 1998, 2002; among many others).
Various grammatical resources have been shown to emerge (Hopper 1987) out of
broader cognitive, discourse, and social-interactional aspects of language use – a
view which has widely come to be known as usage-based approaches to grammar
(cf. Barlow and Kemmer 1999; Bybee 2006; Bybee and Hopper 2001; inter alia).
While a thorough review of this literature lies well outside the scope of the pres-
ent paper, the relevance of this approach becomes clear in light of the ubiquity
of stancetaking discussed above and outlined in the introduction to this volume.
Since stancetaking is such a pervasive activity found in language use, then focus-
ing on stancetaking would seem a natural, worthwhile next step in the devel-
opment of functionalist research paradigms concerning usage-based models of
grammar. Substantial work has yet to be done to flesh out the role of stancetaking
in motivating and constraining grammar, both cross-linguistically, and system-
atically within specific languages. The current paper does not attempt to address
the proposed underlying causal nature of this relationship; rather, I offer here a
preliminary look at how three aspects of the grammar of one particular language
are implicated in stancetaking, as a first step to initiating interest in a functional
link between stancetaking and language form.
As alluded to above, considerable English-centered research already exists that
explores the connection between stancetaking and grammar. Researchers in both
quantitative corpus linguistics and systemic-functional linguistics have worked to
identify macro-level features of lexis and grammar that serve as markers of stance
or evaluation. In particular, the grammar of English modals has proven to be a
rich area for the epistemic evaluation of propositions (cf. review and discussion
in Thompson and Hunston 2000: 20–21), as have epistemic phrases such as I think
and I guess (cf. Kärkkäinen 2003, this volume). English adverbials have also been
widely investigated as signaling various types of epistemic, attitudinal, and style
stances (cf. Biber and Finegan 1988, 1989; Conrad and Biber 2000). And going
beyond the traditional systems of English grammar, Hunston and Sinclair (2000)
have proposed a “local grammar” of evaluative adjectives and nouns. Yet, outside
of English, the link between grammar and stancetaking has virtually gone un-
noticed in descriptive field linguistics. It would be unusual indeed to open to the
table of contents of a reference grammar of another language and find a chapter
devoted to stance, like the one we find in, for example, the Longman Grammar of
Spoken and Written English (Biber et al. 1999) where chapter 12 is entitled, The
grammatical marking of stance.
The purpose of this paper is to explore some aspects of the grammar of stan-
cetaking in a language other than English – in this case colloquial Indonesian –
and to show ways in which traditional grammatical categories described for this
language must also be understood as doing stance work. One of the reasons why
reference grammars have generally not included categories such as stancetaking is

likely due to pervasive (mis)conceptions of grammar as being primarily a cogni-
tively-oriented, propositional system for information coding. But when grammar
is also understood as simultaneously being firmly rooted in social interaction, we
see that much of what has previously been ascribed to meaning and information
coding can, in addition, also be understood as social and interactional in nature.
This paper seeks to flesh out three such resources in colloquial Indonesian – lin-
guistic forms that have general grammatical functions, but, when examined in real
contextualized language-in-interaction, also function at a local level of discourse
to index stance. After a brief discussion of the data used for this study, I will focus
on first-person-singular referring expressions, the -nya clitic, and verbal diathesis
(voice). In conversational interaction, these may function in stancetaking to in-
dex identity, epistemicity, and positioning respectively. I have chosen these three
grammatical resources specifically because they occur frequently in colloquial
Indonesian, represent three very different facets of stancetaking, and clearly il-
lustrate the kinds of gains that can be achieved by incorporating stancetaking into
the realm of grammatical description.
2. Data
The database for this study consists of a corpus of six transcribed audio recordings
of naturally-occurring spontaneous Indonesian discourse. The corpus comprises
36,265 total words in 12,972 Intonation Units (IUs), nearly four hours of running
speech. These six segments are part of larger speech events, which make up the
approximately 25 hours of spoken colloquial Indonesian I collected while con-
ducting fieldwork in Yogyakarta (Central Java) in 1996. The speech events were
not elicited, and I (the foreign researcher) was not present during the recordings.
Four of the segments in the corpus are casual face-to-face conversations among
friends, the fifth is a discussion about music and art by a university student group
that meets monthly, and the sixth is a radio call-in show about Islam. Native-
speaking research assistants and I transcribed the segments using a modified ver-
sion of the Discourse Transcription system outlined in Du Bois et al. (1993), and
subsequently glossed and coded them into a relational database. Englebretson
(2003) provides further details about the six segments in the corpus, and about
the specific methodology used for recording the speech events, choosing the seg-
ments, and transcribing and glossing the data. Appendices A and B of this paper
outline the transcription and glossing conventions. All examples presented in this
paper come from these six speech events, and are labeled with the title of the tran-
script and the line-number(s) of the IU(s) cited in the example.
In addition to the audio files and transcripts, the colloquial Indonesian cor-
pus also includes basic ethnographic data about each of the speakers (with the
exception of those appearing on the radio call-in show for whom this informa-
tion was unavailable). Speakers range in age from late teens to late 20s, and most
are students at various universities in the city of Yogyakarta. The majority of the
speakers (as with the majority of the Indonesian population in general) are from
the island of Java; but the corpus comprises speakers from diverse regions of In-
donesia as well, including Irian Jaya, Sumatra, Sulawesi, and East Timor (which,
at the time of recording, was still part of the Republic of Indonesia). All speakers
claim Indonesian as a native language, were educated in Indonesian, and use it
in their daily lives and interactions, as is typical of educated urban residents of
Indonesia. Most speakers speak a local language as well, typically Javanese, but
report using Indonesian as the language of communication with each other most
of the time – as is represented in the corpus. I have characterized this language
variety elsewhere as spoken colloquial Indonesian (cf. Englebretson 2003), and
noted its similarities with what other researchers have called Spoken Jakarta In-
donesian (cf. Wouk 1989, 1999), which, as noted by Poedjosoedarmo (1982: 142),
has spread widely outside of Jakarta as well.
Before moving on to our discussion of the three aspects of stancetaking in
Indonesian that form the core of this paper, a few brief notes on Indonesian
grammar are in order. There are significant differences between the grammar of
colloquial Indonesian and the grammar of the formal, standardized variety of
Indonesian promulgated by the government, education, and media. For an over-
view of colloquial Indonesian grammar see Ewing (2005), and for an overview of
standard Indonesian see Sneddon (1996) among others. Typologically, Indone-
sian is a primarily isolating language with sparse inflectional morphology. Person,
number, and tense/aspect/mood categories are not indicated on verbs. Verbal af-
fixes consist chiefly of voice and valence marking: agent-trigger, patient-trigger,
middle, reflexive, applicative, and others. In conversation, noun-phrase ellipsis
is quite frequent, so that clauses in fact often consist of a verb alone, and the NP
arguments must be determined based on contextual and pragmatic factors. The
relevance of these basic grammatical facts about Indonesian will become clear as
the discussion progresses throughout the paper.
3. First-person-singular reference as indices of identity
I will begin the discussion of grammatical resources used for specific stance pur-
poses by focusing on the complex system of first-person referring expressions
available to speakers of colloquial Indonesian. While English and many other lan-
guages have only one unique first-person-singular (1SG) pronominal form (e.g.,
English I and its paradigmatically-related inflections me, my, and mine), colloquial
Indonesian has several unique, paradigmatically-unrelated, 1SG forms. The gen-
eral grammatical function of these expressions is, of course, as a means for the
speaker to refer to the self in the ongoing discourse. However, in addition to this
referential function, the variation observed among these multiple 1SG forms as
used by a single speaker in conversation illustrates a specific means of stancetak-
ing. This type of stancetaking is self-expressive, in that it serves to indicate how
the speaker is constructing the self to be perceived by others. For example, just as
culturally-relevant body posture, or physical stance, can index social relations and
attitudes (arms folded across the chest to index social distance from or indifference
toward an interlocutor; a particular hand gesture to signal in-group social affilia-
tion; etc.), self-expressive linguistic stance likewise reflects and creates locally-rel-
evant aspects of social and personal identity. In colloquial Indonesian, the varying
forms available for first-person-singular reference are one resource that speakers
can draw upon to evoke such things as social distance, casualness, or toughness.
As compared with many local languages in Indonesia that have intricate lev-
els of speech styles to index and maintain social relations (e.g., the speech levels
and vocabulary registers of Javanese, as described by Errington 1988), Indonesian
is relatively restricted in the kinds of power and solidarity relations that it allows
speakers to express. This is often noted by Indonesian speakers in language atti-
tude surveys. For example, the following quote is typical of the attitudes of many
bilingual Javanese speakers toward Indonesian, as cited by Errington in his work
on social identity, language shift, and codeswitching in Java.
In response to my various queries about differences between their two languages,
Javanese commonly alluded to Indonesian’s lack of stylistic elaboration as mak-
ing it simple (I: sederhana), bland (I: tawar), or plain (I: polos) in comparison
with Javanese. But, at least as normatively described, Indonesian does incorpo-
rate a few interactionally crucial stylistic distinctions in its personal pronoun rep-
ertoires. (Errington 1998: 92)
And indeed, in Indonesian as in many languages, pronouns are one means used
for stancetaking, to create and maintain social relations and identity. The social-
relational work of pronouns has long been noted for other languages, for example,
the social-psychological work of Brown and Gilman (1960), which characterizes
the “familiar” and “formal” distinction found in the second-person pronouns of
many European languages as reflecting interpersonal solidarity or deference be-
tween speaker and addressee. The so-called t/v distinction (for the French pro-
nouns tu and vous) illustrates that pronouns are far more than merely referring
expressions in the structuralist sense – speakers use them to indicate social rela-
tionships of affiliation or distance. From a more anthropologically-informed per-

spective, Mühlhäusler and Harré (1990) provide an overview of how pronouns
in various languages relate to culturally-diverse concepts of the self, and how
pronouns function in the expression of social relations and attitudes. Colloquial
Indonesian speakers have at their disposal a range of personal pronouns, includ-
ing second-person forms reminiscent of the t/v distinction widely discussed for
other languages. However, what I wish to focus on for purposes of this paper is
the rather unusual case (typologically speaking) of the multiplicity of forms that
Indonesian speakers have available for 1SG reference.3 While grammars and text-
books of Indonesian typically contrast these forms based on macro-level variables
of region and formality, evidence from the corpus demonstrates that these also
function at a local level of discourse, and are used by a single speaker within the
same speech event. Table 1 lists each of the 1SG referring expressions found in the
corpus, along with their overall token frequencies.
This table shows, in descending order of frequency, the five forms that speak-
ers in the corpus regularly use in referring to themselves. Because of the high
prevalence of argument ellipsis in colloquial Indonesian conversation, a sixth op-
tion also available to speakers is to use no form at all – so-called argument ellip-
sis – thus leaving the identity of the unexpressed arguments to contextual infer-
ence. Many of the clauses in the corpus, therefore, have no overt 1SG form, even
though they are clearly referring to the speaker. The question of what motivates
a speaker to either use or to ellipt a 1SG form is of great interest in its own right,
but lies outside the scope of the present paper.4 I am concerned here with the ways
in which overt first-person-singular forms are used, and, in particular, how they
function in stancetaking. I will leave the issue of colloquial Indonesian argument
expression versus ellipsis for future research, focusing here only on the five overt
forms shown in the preceding table.
The first four (aku, sya, gua, and tak) are 1SG pronouns, and the fifth is self-
reference through the use of one’s own name. In all discussion of these forms in
the literature that I am aware of, including textbooks, dictionaries, pedagogical
materials, and descriptive grammars, these 1SG forms are characterized in terms
Table 1. Total 1SG referring expressions in the corpus

Form Total tokens
aku (aku, ku-, and -ku) 578
saya 222
gua/gue 83
tak 37
proper name 35
of the traditional macro-level sociolinguistic variables of formality, region, and

ethnicity as follows. The most frequent of these pronouns observed in the corpus
is aku, including its paradigmatically-related clitic forms ku- (a first-person pro-
clitic on verbs) and -ku (a possessive enclitic and object of a preposition). Aku is
generally characterized as familiar and informal, used among intimates, or by a
person of higher social status to a person of lower social status. The second-most
frequent 1SG pronoun in the corpus is saya, which is prescriptively the all-pur-
pose Indonesian 1SG pronoun. It is considered more formal than aku, and the
most appropriate form for public interaction and among people who are not so-
cially intimate. The next-most-frequent form in the corpus, gua, and its variant
pronunciation gue, was borrowed into Indonesian through Hokkien Chinese. It
is typically associated with young urban residents of Jakarta and/or speakers of
Chinese-Malay background. The least-frequent 1SG pronoun in the corpus is tak,
which is generally characterized as a Javanese loan. The final entry in the table
indicates that there are 35 instances in the corpus of speakers referring to them-
selves by use of their own name. According to the impressions of native-speaking
Indonesian consultants, the use of one’s own name is generally restricted to young
unmarried women – and this observation holds true for all 35 instances in the
corpus. The following entries from Echols and Shadily’s (1989) Indonesian-Eng-
lish dictionary summarize the 1SG pronoun forms found in my corpus, reflecting
the way these pronouns are generally treated in the Indonesianist literature.
aku “I (familiar, intimate)”
saya “I, me, my”
gua “(Jakartan, Chinese, Colloquial) I, familiar first person pronoun”
tak “(Javanese) I, by me”
Readers who are not familiar with Indonesian grammar may wonder whether any
of these forms are inflectionally related to each other. The answer is negative. Each
is invariant for grammatical categories such as case or possession; other than the
clitic forms of aku, no paradigmatic forms of these pronouns exist. In sum, two of
these forms (aku and saya) are traditionally characterized by differences in for-
mality, the remaining two pronouns (gua/gue and tak) are characterized based on
the regional/ethnic/linguistic background of the speaker, and the use of self-refer-
ence by one’s own name is characterized in terms of the gender and marital status
of the speaker. Such definitions and general characterizations have tended to be
based on impressionistic, intuition-based claims of analysts and native speakers,
and, as I will demonstrate shortly, do not fully account for their use in conversa-
tional interaction. There has been an overall lack of empirical research into the
actual distribution of these forms in naturally-occurring language-in-use. A wel-
come exception is Sneddon (2002) who provides a quantitative analysis of 1SG
forms (and other stylistically-stratified variables) in a corpus of various genres of

spoken Indonesian among residents of Jakarta. Sneddon’s quantitative findings do
bear out the claims that 1SG pronouns vary based on formality level. Results from
my own corpus support this as well, based on the following evidence.
As mentioned in Section 2, above, five of the speech events in my corpus
consist of face-to-face conversation among people who know each other well,
while the sixth speech event (which I have entitled Tanya-Jawab ‘question and
answer’) is an entirely different genre of spoken language – a phone-in radio show
in which a radio host and a Muslim cleric answer callers’ questions about Islam.
This speech event is different from the other five in a number of ways: (1) it is
a public, radio-broadcast telephone interaction, while the other five are private,
face-to-face conversations; (2) the callers are not socially acquainted with the ra-
dio host or the cleric, whereas in the five other speech events, participants know
each other well and interact with each other regularly; (3) in the call-in show,
participant roles and social relations are clearly delineated – the cleric is in a posi-
tion of authority as expert, and the callers are positioning themselves as lacking
the knowledge that the cleric provides; in the other five speech events, participant
roles are more balanced, and there is no overtly recognized authority or expert.
On all three of these dimensions, then, Tanya-Jawab represents a far more formal
register than any of the other five speech events in the corpus. Thus, if 1SG pro-
nominal forms are indeed sensitive to formality level, then we would expect to
find significant differences between the radio call-in show as contrasted with the
other five informal face-to-face conversations. As shown in Table 2, the data do
indeed provide strong evidence for this claim.
This table shows a striking difference between 1SG reference in the Tanya-
Jawab speech event as compared with the aggregate use of 1SG reference in the
other five.5 In the radio call-in show, speakers overwhelmingly prefer saya, with
98 instances of this form and only one instance of any other (self-reference by
means of the speaker’s own proper name). In this speech event, there are no in-
stances whatsoever of the familiar aku nor of the regionally/ethnically marked
gua/gue or tak. These numbers provide strong evidence that speakers treat saya
as the 1SG pronominal form most appropriate for formal public interaction. In
contrast, the overwhelming preference for aku found in the conversational speech
events suggests that speakers treat aku as the preferred form for informal inter-
action among social familiars. These empirical findings seem to provide robust
support for the general claims that the macro-level variables of formality and in-
timacy strongly determine the use of saya or aku respectively. However, as I will
demonstrate in the remainder of this section, the situation is far more complex
than could be accounted for by a direct deterministic relationship between social
variables and linguistic form.
Table 2. Comparison of 1SG reference in informal and formal speech events
Form Face-to-face tokens “Tanya-Jawab” tokens Total
aku 578 0 578
saya 124 98 222
gua/gue 83 0 83
tak 37 0 37
proper name 34 1 35
The data suggest that pronoun choice in conversational interaction is not pre-
determined based on static social variables, but is dynamic, takes place at the local
level of discourse, and is used in stancetaking to index the speaker’s construction
and expression of identities. Primary evidence for this claim comes from observ-
ing 1SG usage by single speakers, with the same interlocutors, in the same speech
event. In the five face-to-face conversations in the corpus, there are no speakers
who refer to themselves with only one 1SG form; every speaker in the corpus
(aside from those in the Tanya-Jawab radio call-in show) regularly uses more
than one 1SG form. Following is a table illustrating the use of 1SG forms in the
speech of Ari (speaker A) from the Pencuri speech event. Ari is one of the more
loquacious speakers in the corpus, 21 years of age, female, originally from Eastern
Indonesia, and a student in Yogyakarta at the time of recording. Note that her
repertoire of 1SG forms includes all five found in the corpus.
Table 3 shows a single speaker using all five of the 1SG forms to refer to her-
self, with the same interlocutors, in the same physical setting, during the same
speech event. I have produced similar tables for all speakers in the corpus, and as
mentioned above, there are no single-form speakers in the face-to-face conversa-
tions. Not all speakers use all five forms in such striking numbers as does this
particular speaker, but it is an empirical fact that every speaker in the face-to-face
conversational data uses two or more. These findings suggest that the traditional,
top-down, a priori account of what motivates 1SG use is incomplete.
As an actual example of the use of 1SG forms in real-time conversational
interaction, I offer the following short excerpt of 18 intonation units (lasting
17.7 seconds) of a conversational narrative, again from Ari in the Pencuri speech
event. This excerpt is part of a larger narrative Ari is telling about a friend who
had allegedly stolen money from her and then lied to cover up the theft. For the
benefit of non-Indonesianist readers and to facilitate the reading of this excerpt, I
have included the actual Indonesian 1SG forms in parentheses in the English free
translation. Other material in parentheses indicates that no corresponding form
is used in the Indonesian data, but is necessary in the English free translation.
Table 3. 1SG forms in the speech of a single speaker

Form Tokens
aku 81
saya 40
proper name 25
gua/gue 8
tak 5
(1) Pencuri IU 209–227

209 A: ... Dia di ini.
3sg at this
A: She was at, um,
210 .. Di depan pintu rumah Ari.
at front door house Ari
in front of the door to my_(Ari) house.
211 D: Kamu tanya dia.
2sg ask 3sg
D: You asked her?
212 A: .. Aku kan,
1sg prt
A: I_(aku) was um,
213 bingung lho kok.
confused prt prt
(I) was confused, you know.
214 ... Perasaan tadi tuh,
feeling just:now prt
(My) feeling just then was that,
215 .. yang pegang dompet dia kan,
rel hold wallet 3sg prt
she was the one who had taken (my) wallet.
216 [terus gua bilang-in],
next 1sg say-app
[So then I_(gua) said to her],
217 M: [Yah].
yeah
M: [Yeah].
218 A: ... Nama-nya kan Rifka.

name-pos prt Rifka
A: Her name was Rifka.
219 ... Eh Rif,
hey Rif
“Hey, Rif!
220 .. Kamu,
2sg
You,
221 tadi nge-lihat-in,
just:now at-see-app
were just now looking at,
222 duit saya yang dalam= dompet ya.
money 1sg rel inside wallet prt
my_(saya) money that was inside (my) wallet, weren’t you?”
223 Nggak tuh,
neg prt
“No, hunh-unh.
224 nggak tahu.
neg know
(I) don’t know (anything about it).”
225 ... Terus Ari masuk lagi cari ke dalam,
... next Ari enter again search to inside
So then I_(Ari) went in again and looked for it inside.
226 dia itu ngak- --
3sg that neg
She didn’t –
227 nggak lari gitu lho.
neg run like:that prt
[she] didn’t run away.
Note that in this brief excerpt of only 18 IUs of conversation, a total of 17.7 seconds
of talk, the speaker seamlessly switches among four of the five 1SG forms found in
the corpus. She uses Ari (her own name) in IUs 210 and 225, aku in IU 212, gua
in 216, and saya to refer to herself in reported speech in IU 222. I will now turn to
a discussion of two possible hypotheses to explain the 1SG variation seen in this
brief excerpt and in the corpus more broadly: (1) distribution of 1SG forms could
be grammatically conditioned, and (2) distribution of 1SG forms could be based
on macro-level social variables. Both of these hypotheses fail to fully account for
what we actually observe speakers doing, and I will therefore suggest a third ap-
proach which more thoroughly accounts for the observed distribution in 1SG
forms; namely, speakers are actively using 1SG forms in stancetaking in order to
construct and project aspects of their social and personal identities.
The first hypothesis to consider is that 1SG forms may be motivated and con-
strained by grammar. If a speaker’s choice of 1SG form were determined by gram-
matical factors, then, following classic structuralist reasoning, we would expect to
find 1SG forms occurring in some sort of contrastive distribution. In other words,
some 1SG forms in the corpus would occur in grammatical environments where
other 1SG forms do not occur. However, (with the possible exception of tak, to
which I will return shortly), there appear to be no such grammatical restrictions.
For example, the 1SG form saya in IU 222 duit saya ‘my money’ would be per-
fectly acceptable with the substitution of any of the other 1SG forms found in this
short excerpt: duit gua (using gua/gue), duitku (using the enclitic form of aku),
or duit Ari (using the speaker’s own name) are all acceptable and natural. In fact,
these very tokens are actually attested elsewhere in the corpus, as illustrated in the
following three examples.
(2) Pencuri IU 2403

Duit -ku nggak ada.
money -1sg neg ex
I don’t have any money. (lit., ‘My_(aku) money doesn’t exist.’)
(3) Dingdong IU 1002

Nggak ada duit gua.
neg ex money 1sg
I don’t have any money. (lit., ‘My_gua money doesn’t exist.’6)
(4) Pencuri IU 266

Cabut duit Ari semua.
pull money Ari all
(She) pulled out all my_(Ari) money.
These three examples, along with IU 222 in excerpt (1), above, illustrate that each
of these four 1SG forms (saya, aku, gua, and self-reference with proper name) all
occur in the same grammatical environment: as a 1SG possessor in a possessive
NP headed by duit ‘money.’ These four forms, along with tak, are all fully attested
in the corpus in other grammatical roles too: as both the trigger and the non-
trigger argument in transitive clauses, and as the single argument of intransitive
clauses. These distributional facts suggest that the variation among 1SG forms is
not grammatically conditioned.
The one form which does appear to have some grammatical restriction is tak;
it never occurs in the corpus as a possessor in a possessive NP, and it tends to
strongly collocate with certain verbs of cognition. Eighteen out of the 37 tokens
of tak in the corpus co-occur with the verb pikir ‘think,’ suggesting that tak pikir
‘I think’ is a grammaticized epistemic phrase (cf. Kärkkäinen 2003, this volume
on English I think and I guess respectively; cf. Rauniomaa, this volume, on similar
constructions in Finnish). However, while this collocation does suggest a restrict-
ed grammatical profile for tak, it nonetheless overlaps with the other 1SG forms in
other environments. The hypothesis that the variation of 1SG forms is grammati-
cally conditioned receives minimal support from tak, and no support whatsoever
from the other four 1SG forms.
The second hypothesis to consider is that 1SG use is linked to macro-level
social and stylistic variables such as formality and regional background. I have al-
ready discussed this to some extent above, and have shown, based on a comparison
of 1SG usage in a radio call-in show with 1SG usage found in everyday conversa-
tion among social familiars, that saya and aku are strongly correlated with formal-
ity. Saya is the only 1SG form found in the call-in show (except for one instance
of self-reference by proper name), while aku is the preferred form in the conversa-
tional data. However, this cannot account for the fact that there are still 124 tokens
of saya in the informal conversational data, as shown in Table 2. It also cannot
account for the distribution observed in Table 3, where a single speaker freely uses
both saya and aku (along with the other 1SG forms) in the same speech event, and
even within the same short span of discourse as seen in excerpt (1).
The regional account of gua and tak is similarly problematic. For example, as
shown in Table 3, Ari uses eight tokens of gua (which is claimed to be colloquial
and Jakartan) and five tokens of tak (which is claimed to be Javanese), along with
the other three 1SG forms. However, this speaker has never lived in Jakarta at all
(thus suggesting that calling gua Jakartan is an oversimplification), and claims not
to speak Javanese (suggesting that tak is not simply Javanese code-mixing.). Thus,
a regional account of gua and tak does not account for much of what speakers are
actually doing in conversational interaction.
The social variables hypothesis does, however, appear to receive strong sup-
port from 1SG reference by means of proper name. Traditionally, this is seen as
a form used by young, unmarried women; and the three speakers in the corpus
who use this form are in fact young, unmarried, and female. However, there are
unmarried women of the same age in the corpus who never use this form, and the
speakers who do use their names as 1SG reference also use other 1SG forms as
well (e.g., Ari in Table 3, who uses all five 1SG forms). So for speakers like Ari who
do use their names as 1SG reference, the social variables hypothesis still does not
explain what would condition her to use her name at some points in the conversa-
tion, and to use different 1SG pronouns at other points in the talk.
An analysis of 1SG usage based on social variables fails to account at all for
the observation that every speaker in the conversational data uses two or more
forms. As shown in Table 3, a single speaker may use all five of the 1SG forms,
despite the fact that she is using them all in the very same speech event. In other
words, the variables of formality and regional origin for Ari are remaining con-
stant, yet her use of 1SG form varies; she is speaking with the same interlocutors
(housemates and friends whom she has known for nearly two years), in the same
physical context (in the common room of their boarding house), and her ethnic/
regional/language background also does not change throughout the interaction.
The fact that 1SG usage varies, while the macro-level sociolinguistic variables re-
main constant, suggest that there is an additional level of detail not captured by
the social variables hypothesis.
These shortcomings of the social variables hypothesis reflect larger, prob-
lematic assumptions present in much of traditional variationist sociolinguistic
research regarding the nature of identity categories in general. Variationist so-
ciolinguistics has tended to assume identity “as a pre-discursive construct that
correlates with, or even causes particular [language] behaviours” (Benwell and
Stokoe 2006: 26). In other words, in traditional variationist work, social categories
of identity are treated as relatively stable, a priori characteristics of the speaker
(age, gender, regional origin, language background, etc.) or stable attributes of the
particular speech situation (e.g., formality, casualness, etc.) In this view, the par-
ticular variant a speaker uses in any given instant of talk falls out probabilistically
from these pre-existing social and stylistic categories. If, on the other hand, we
understand identity categories (such as gender, age, regional origin of speakers,
or formality of speech events) from a performative point of view – as something
speakers ‘do,’ rather than as something speakers ‘are’ – then we can begin to ap-
proach the variation among Indonesian 1SG forms from a new perspective, one
which stands a much better chance of accounting for how speakers use these forms
in everyday talk than do either of the hypotheses examined above. Rather than
seeking macro-level variables to explain the distribution of 1SG forms in the data,
a performative approach would ask: What is it that a speaker is accomplishing by
using this particular 1SG form at this particular moment in the interaction?, i.e.,
What kind of stance is the speaker evoking by using this form at this time?
According to a performative conception of identity, speakers may use lan-
guage to evoke social or personal identity categories in order to achieve certain
goals within talk. Rather than being relatively stable attributes that pre-exist the
current discourse, identity categories are emergent, socially constructed practices
of language use (cf. Bucholtz and Hall 2005 for similar discussion). The many types
of identities in discourse include “local identity categories and transitory inter-
actional positions” and “temporary and interactionally specific stances and par-
ticipant roles” (Bucholtz and Hall 2005: 592). Accordingly, one way that a speaker
“performs” identity is through the packaging of utterances in ways that reflect and
index specific categories. As demonstrated in Table 3, and in excerpt (1), the speak-
er has several potential options at any given moment in talk, as to how to package
the utterance for 1SG self-reference. I shall now offer some qualitative observations
of how 1SG forms are used in the performance of identity. I suggest that these vari-
ous forms serve to index, or evoke, specific types of identity categories.
Bearing in mind a performative conception of identity, I will begin by dis-
cussing the use of aku and saya from this perspective. The distribution in Table
2 has already demonstrated that there is a strong correlation between these two
pronouns and the types of speech events in which they occur. Saya is the only
1SG pronoun used in the call-in radio show, with 98 tokens (plus one 1SG refer-
ence by use of proper name). I suggest that speakers here are actively using saya,
among other lexical and grammatical items, to take a formal stance: to construct a
public social and personal identity, to show deference, and to index social distance
among participants. On the other hand, in the casual conversational data, aku is
overwhelmingly the most frequent form, with 578 tokens in these five speech
events – nearly five times more frequent than the use of saya in these same speech
events. I would contend that speakers here are using aku as one means of taking
a casual stance: actively constructing their personal identities as informal, relax-
ing the prescribed norms of public language use, and building the social intimacy
among speakers typical of this type of interaction.
But what about the 124 tokens of saya in the conversational data? If speakers
are actively constructing these interactions as casual and intimate, as suggested
by the high frequency of aku, then why are these same speakers simultaneously
using saya, with its connotations of formality and distance? I suggest that when
speakers use saya in informal conversation, they are shifting their stance: they are
using this form to construct social distance or formality at a particular point in
the ongoing interaction. (To return to the analogy of physical stance, the use of
saya in informal conversation could be compared with the physical act of tensing
the body or temporarily adopting a more rigid posture.) A crucial observation
which lends support to this idea is that most of the tokens of saya come from
instances of reported speech or thought (which, following Tannen (1986, 1989)
I shall hereafter refer to as constructed dialogue), even when a different pronoun
would have been used in the original context of utterance. As an example, con-
sider the use of saya in IU 222 of excerpt (1), above. In this clause, which starts
in IU 216 with the quotative verb bilang ‘say,’ Ari is reporting the words she alleg-
edly had spoken to her friend, “Hey, Rif! You were just now looking at my_(saya)
money that was inside (my) wallet, weren’t you?”. Interestingly, if 1SG usage were
conditioned by casualness and intimacy, one would have expected aku here, since
aku would most certainly have been the form she would have used with her friend
in the original context. The form aku also would have been in parallel with the
second-person familiar pronoun kamu with which she addresses her friend in the
constructed dialogue found in IU 220. However, the speaker is using saya, pre-
sumably to reflect a more distant stance than the informal aku. For several speak-
ers in the corpus, saya is only found in contexts of constructed dialogue – either
the alleged words of others, or in the speakers own reported utterance as in this
example. Based on these observations, I suggest that speakers are using saya to
take a stance on reported speech or thought: to imply social distance from it.
While saya is the norm in constructed dialogue in the corpus, it is by no
means the only form of 1SG reference in this context; aku and other 1SG forms
are commonly found in reported speech and thought as well. For this reason, it is
illuminating to compare the example of saya in IU 222 of excerpt (1) (as just dis-
cussed) with another instance, one in which the speaker uses the more intimate
aku. The following excerpt comes from a narrative about a woman named Maya
who is nearly tricked into buying shoddy towels at an outdoor market. The full
narrative will be presented and explicated for other purposes in Section 5 below,
but for the present discussion, the relevant aspect of this example is the use of aku
in the constructed dialogue in IU 2753.
(5) Pencuri IU 2751–2757

2751 I: .. Maya itu udah ter-tarik banget.
Maya that perfv nonvol-pull very
I: Maya got really attracted.
2752 Ah murah ini,
oh cheap this
“Oh, this is cheap!
2753 aku pingin handuk tiga ribu itu.
1sg want towel three thousand that
I_(aku) want this towel for 3,000 (Rupiah).
2754 Padahal gede tebal itu,
in:fact big thick that
It’s big and thick.
2755 harus‑nya enam ribu lebih to.
should‑stm six thousand more prt
It should be 6,000 (Rupiah) more.”
2756 dia kan,

3sg prt
he was (like),
2757 .. mikir‑nya ah lumayan @mahasiswa @baru @gitu ya.
thought‑pos prt reasonable student new thus prt
Her thoughts were reasonable for a new student, you know.
In IU 2751, the speaker (Indra) provides an abstract of this portion of her narra-
tive, in which she summarizes what happens to Maya at the market: “Maya got re-
ally attracted” (by the towels she saw ). The subsequent four IUs construct Maya’s
inner dialogue, including her thoughts about the towels and their price, and IUs
2756–2757 explicitly frame this speech as having been constructed dialogue “her
thoughts…”. The IU of interest for the present discussion is IU 2753, where Maya is
reported to think: “I_(aku) want this towel for 3,000 (Rupiah).” What is notewor-
thy about the constructed dialogue in IU 2753 of this excerpt, as contrasted with
Ari’s constructed dialogue in IU 222 of excerpt (1) above, is that Maya’s reported
‘thoughts’ use the more intimate aku, while Ari’s reported ‘speech’ uses the more
distant saya. If one difference between saya and aku in constructed dialogue has
to do with the distance/intimacy or degree of vividness which the speaker is cre-
ating, then this observed use of saya for ‘speech’ and aku for ‘thoughts’ is indeed
motivated by the speaker’s social world being constructed through stancetaking;
perhaps ‘thoughts’ are being presented as more internalized to the speaker, and
are thus constructed with greater intimacy, while ‘speech’ is more observable and
less internalized, and thus is constructed with greater distance (cf. the theory of
“territory of information”, as articulated by Kamio 1997).
While a more thorough study of constructed dialogue in Indonesian con-
versation is clearly warranted to assess this conjecture, I would like to suggest
that the use of saya versus aku in these contexts may be similar to the effect that
many languages achieve by a distinction between indirect and direct speech. In
other words, the use in constructed dialogue of the less-intimate form saya versus
the more-intimate form aku may be one means that Indonesian speakers have of
marking it as less vivid, as lower in epistemic value, and (for whatever reason) as
more distant from the speaker. On the other hand, the use of aku in constructed
dialogue reflects a conception of it as more vivid, as higher in epistemic value,
and as less distant. Clearly this observation warrants further in-depth investiga-
tion, but I offer it here along with the presented examples to illustrate one resource
which colloquial Indonesian speakers have of taking a stance. In sum, in informal
colloquial Indonesian conversation (whether in the context of constructed dia-
logue or not), the use of saya or aku is one means of stylistic elaboration avail-
able to speakers, as they use language to construct their social worlds through
stancetaking.
The third-most frequent 1SG form in the corpus is gua/gue, with 83 tokens
in the informal conversational data. This pronoun is generally associated with
colloquial Indonesian spoken in Jakarta and/or by speakers of Chinese ethnicity.
Yet, as discussed earlier, a regional/ethnic characterization of this pronoun does
not accurately account for the observed data. For example, Table 3 shows that Ari
uses eight tokens of gua in the corpus, yet she has never lived in Jakarta and is
not ethnically Chinese. Just as with saya and aku, I suggest that speakers use gua
in stancetaking, to evoke some type of temporarily relevant identity in the ongo-
ing discourse. As a first approximation of what such a temporarily-relevant iden-
tity might be, consider Ari’s use of gua in IU 216 of excerpt (1) above: “So then
I_(gua) said to her…” Note, crucially, that this form occurs to frame the rather
blunt accusation Ari reports having made of her friend (IUs 219–222): “Hey, Rif!
You were just now looking at my money that was inside my wallet, weren’t you?”
In this short excerpt, Ari presents herself as blunt and outspoken, constructing
her identity as someone who is ‘tough’ and unafraid to confront her friend whom
she believes is a thief. I would like to suggest that her use of gua contributes to
this locally-relevant construction of identity. Rather than indexing regional af-
filiation per se, gua draws upon common language attitudes and stereotypes of-
ten associated with speakers who use this form. Responses of several Indonesian
consultants to excerpts from this corpus where gua is used bolster my claim that
these stereotypes are related to ideologies of being “tough” and “outspoken.” I
readily acknowledge the need for both more in-depth language attitude surveys
and qualitative sequential analysis of conversation to assess this hypothesis, but as
a preliminary step toward this goal, I offer the case of the most prolific user of gua
in the corpus: speaker L from the speech event entitled Dingdong (the onomato-
poeic name of a pinball arcade game discussed in the speech event). This speaker
is male, 19 years of age, and had lived in Jakarta from age 10 until age 13. He uses
only two 1SG forms in the data: 63 tokens of gua and seven tokens of aku. In other
words, of the 83 gua tokens in the data, speaker L produces 63 of them (75.9%).
When I played portions of this recording separately to three Indonesian consul-
tants, all expressed the view that this speaker sounded ‘tough’ or ‘rude’ or ‘macho,’
and one consultant was offended by the topics of this spontaneous conversation
(playing video games, drinking at a local bar, and flirting with women). I suggest
that one of the ways in which speaker L is able to construct an identity of being
‘tough’ and ‘outspoken’ is through his use of gua. When speakers use gua, they are
indexing these stereotypes, whether or not they have any affiliation with Jakarta
at all (cf. Rampton 2005 for work on “crossing,” or appropriating the speech styles
of other groups for specific social purposes; Johnstone, this volume, on the use of
stancetaking to bolster specific local identities).
To summarize this section, I have shown that 1SG self-reference in collo-
quial Indonesian is an interactional practice available to speakers for creating,
maintaining, and implying facets of their social and personal identities. I readily
acknowledge that substantial work remains to be done on this topic, both quanti-
tative and ethnographic. I have not offered an account here of how tak or proper
names are used in stancetaking, partly because there are so few tokens of these
forms in the corpus (37 tokens of tak and 35 of proper name); I leave this to future
research, with the suggestion that, as with the other three pronouns discussed
here, their stance functions can best be accounted for by observing the kinds of
identity categories speakers evoke at a local level of conversation. In this section,
I have offered here a first step toward the goal of moving Indonesian grammatical
description beyond traditional formal and referential boundaries, and toward a
greater recognition of the ways in which grammatical resources are used in the
service of stancetaking. In this case, 1SG forms do far more than simply refer to
the speaker, and the variation among these forms does not simply fall out from
social variables. Rather, much of self-reference is closely linked to self-expressive
stancetaking, and is managed locally in the ongoing discursive construction of
identity. Just as a speaker’s physical stance reflects and indexes temporary social
categories, the linguistic stance evoked by the use of one 1SG form over another
creates similar transitory identities, such as social distance with saya or tough-
ness with gua. The 1SG forms of Indonesian are a general grammatical resource
which, in addition to their referential function of referring to the speaker, are
simultaneously employed in the construction and maintenance of stance.
4. Epistemic -nya constructions
The previous section has illustrated how Indonesian speakers may use first-per-
son singular pronouns to construct aspects of their personal and social identities.
This type of linguistic stancetaking is self-expressive and reflexive. Just as the act
of physically taking a stance may often entail the outward presentation of some
facet of the self, through body posture or gesture or other means, the linguistic
expression of stance may likewise entail the display of some currently-relevant
part of the speaker’s identity. The kind of stancetaking discussed in the present
section, on the other hand, involves epistemic stance toward something external
to the speaker – in this case toward the current utterance itself. Following Wu, this
type of epistemic stance refers to “a speaker’s indication of how he or she knows
about, is commenting on, or is taking an affective or other position toward the

person or matter being addressed” (2004: 3). In this section, I will illustrate how
Indonesian speakers use the -nya clitic to do just this. In addition to its general,
well-described, referential functions as a marker of third-person possession, defi-
niteness, and nominalization, this clitic frequently occurs on lexemes related to
cognition, utterance, or modality. In such a context, these constructions function
as adverbials, and are used for epistemic stancetaking regarding the current ut-
terance. The kinds of epistemic stances implicated by -nya constructions include:
(1) evidentiality: the source of knowledge of the current utterance; (2) assessment
of interactional relevance: the degree of value a speaker places on the utterance,
usually regarding its role in the ongoing interaction; or (3) affect: the speaker’s
mental or emotional attitude toward the proposition expressed by the rest of the
clause. These stance-related functions have generally gone unnoticed in descrip-
tive grammars of Indonesian, which tend to focus only on the referential and
grammatical aspects of this clitic, with little, if any, discussion of its use in stan-
cetaking.
The -nya clitic is the most prevalent bound morpheme found in the corpus,
with 1,570 tokens in the six speech events. This frequency is greater than the
occurrences of all other grammatical morphemes in the corpus, including the
agent- and patient-trigger verbal prefixes as well as pronominal clitics, suggesting
that -nya deserves recognition as the most prevalent bound morpheme found in
colloquial Indonesian in general. Yet, outside of my own work on -nya (see Engle-
bretson 2003: Chap. 5 for a more thorough discussion than is possible in this brief
section), it has received relatively little analysis. Grammars of formal varieties of
Indonesian typically gloss this as a third-person-singular possessive marker, and
also recognize its use as a marker of definiteness and as a nominalizer. These three
referential uses of -nya are found in the colloquial corpus too, as illustrated by the
following three examples.
(6) Pencuri IU 218

A: ... Nama-nya kan Rifka.
name-pos prt Rifka
Her name was Rifka.
In this example, the -nya clitic occurs on the noun nama ‘name’ to indicate a
third-singular possessor ‘her name.’ The following example contains two tokens
of -nya. The first is a definite marker, while the second again indexes a third-per-
son possessor as in the previous example.

626 D: dari belakang ini,
from back this
627 pencopet-nya buka kalung-nya.
pickpocket-def open necklace-pos
From behind, the thief undid her (Agnes’s) necklace.
IU 627 contains two instances of the -nya clitic. The first marks the noun pencopet
‘thief ’ as definite; this noun was mentioned previously in the discourse context,
in IU 599 (not shown here), and plays a central role in the ongoing narrative. The
second instance of -nya in IU 627 marks third-person possession on the noun
kalung ‘necklace,’ just as in example (6). IU 834 of the next example illustrates a
third grammatical function of ‑nya, that of nominalization.
(8) Dingdong IU 833–835

833 ... pokok-nya bukan minum bir aja.
main-stm neg drink beer just
834 ... Minum-nya,
drink-nom
835 .. me-lebih-i= Martini kata-nya.
at-more-app martini word-stm
the thing is, (they) weren’t just drinking beer.
(Their) drinking even surpassed martinis, they said.
In this excerpt, speaker L is telling his friends about a group of women he had
seen the previous evening at a bar, and is commenting on his shock about the
quantity of alcohol and types of drinks they were consuming. There are three in-
stances of -nya in this example. The first and third are stance markers (STM) to
which we will return shortly. The second, on the verb minum ‘drink’ in IU 834, is a
nominalizer. It marks this word as a noun, which then becomes the subject of the
clause in IU 835, ‘(Their) drinking even surpassed martinis, they said.’
The three functions of -nya we have observed so far – third-person-singular
possession, definiteness, and nominalization – are the typical grammatical func-
tions of this morpheme discussed in descriptive and pedagogical grammars of
Indonesian. Yet, this account of -nya does not fully address the range of use ob-
served in colloquial Indonesian. More than one-third of the instances of -nya in
the corpus, 580 out of 1,570 tokens, are not marking third-person-singular pos-
session, are not indicating definiteness, and are not nominalizing a verb to be a
clausal argument. Rather, these instances of -nya are being used in stancetaking,
enabling a speaker to comment on some aspect of the current utterance: how the
speaker knows about it (evidentiality); how the speaker evaluates it in light of the
current conversation (assessment of interactional relevance); or how the speaker
feels about it (emotional attitude). Grammatically, these expressions are adverbi-
als, as seen from their syntactic freedom to occur at numerous positions in the
clause.7 To illustrate two of the stance-marking functions of -nya constructions,
return to example (8) above. Note that there are three instances of -nya in this ex-
ample, one in each IU. The token found in IU 834 has already been discussed as a
nominalizer, changing the verb minum ‘drink’ into a noun which functions as the
subject of the clause. The remaining two -nya tokens in this example are stance
markers. The first, in IU 833, consists of -nya on pokok (a word roughly translated
as ‘main,’ ‘fundamental,’ or ‘basic’). This serves as an epistemic evaluation of the
interactional relevance of the utterance for the discourse at hand. With this word,
L highlights the rest of the clause in terms of his assessment of its relevance for
the interaction – as something especially important for the interlocutors to pay
attention to. It is the fact that these women were “not just drinking beer,” which
provides essential background for the rest of his unfolding narrative about the
antics of this group of friends. The second instance of ‑nya as a stance marker,
in example (8), occurs in IU 835 on the noun kata ‘word.’ Here, -nya marks evi-
dentiality, the source of knowledge for this utterance, indicating that it originated
from the words of others. That is, speaker L knows what the women were drink-
ing because this is something they had told him. In sum, these two tokens of -nya
illustrate that when this clitic occurs on lexemes related to utterance, cognition, or
modality, it forms an adverbial which marks the speaker’s stance toward the rest
of the clause. The remainder of this section will take up each of the three stance
functions of -nya constructions in turn.
One stance-related function of -nya constructions is the marking of eviden-
tiality, the source of knowledge of the speaker’s utterance. Following are three
more examples, each illustrating a type of evidentiality: knowledge based on the
words of another (example 9), knowledge from general inference (example 10),
and knowledge from visual perception (example 11).

986 ... Dia kan kata-nya mau n-yari saudara-nya
3sg prt word-stm want at-look:for relative-pos
987 di sini.
at here
He said he wanted to visit his relatives here.
This example is parallel to the token already discussed in IU 835 of example (8). In
IU 986 of example (9), kata ‘word’ is suffixed with -nya, again indicating that the
speaker’s source of knowledge is the words of another. The speaker knows that the
person in question wanted to visit his brother, because this was something told
to her by a friend. There are 59 instances of katanya (‘word’-nya) in the corpus,
all of which mark evidentiality based on another’s speech. The following example
illustrates -nya marking evidentiality based on general inference.
(10) Pencuri IU 2299

mati kayak-nya jam Mega.
dead like-stm watch Mega
My watch seems to have stopped.
In this example, the colloquial preposition kayak ‘like/as’ is suffixed with the clit-
ic -nya, making it into an adverbial expressing a general evidential source. The
speaker, Mega, notices that her watch is showing the same time as it had given
earlier, and infers that it therefore must have stopped. Kayaknya (‘like/as’-nya)
occurs 76 times in the corpus. As in this example, this expression does not specify
a particular evidential source, but indicates that the speaker is making the state-
ment based on general inference. IU 1249 of the following example illustrates a
third type of evidential source, in this case visual perception.
(11) Pencuri IU 1247–1251

1247 Trus ada ibu-ibu yang keren itu lho
next ex mother-mother rel well:dressed that prt
1248 .. Pokok-nya,
main-stm
1249 .. ke-lihat-an-nya orang kaya itu lho
nonvol-see-nom-stm person rich that prt
1250 dari <opo>?
from what
1251 .. Dandan-nya,
attire-pos
And then there were some well-dressed women. The thing is, they looked
like rich people – because of, what’s-it?, the way they were dressed.
In IU 1249, the speaker states that the women must be rich, and indicates her
evidential source for this claim as based on how they looked. The verb lihat ‘see’
is affixed with the stative/nonvolitional circumfix ke- -an, changing the meaning
of the verb to ‘visible,’ which is then suffixed with the evidential -nya. In sum, we
have seen three types of evidential sources indicated by -nya constructions: evi-
dence from the words of another, evidence from general inference, and evidence
from direct (visual) perception. This clitic also commonly occurs on the verbs
pikir ‘think’ and rasa ‘feel,’ to indicate the evidential source as one’s own thoughts
and feelings.
Note that example (11) contains two additional tokens of -nya. The occur-
rence of -nya in IU 1251 is a possessive.8 The remaining -nya token, in IU 1248,
is a stance marker, indicating the speaker’s assessment of interactional relevance,
which is the second type of epistemic stance we will be addressing in this sec-
tion. As discussed previously regarding IU 833 in example (8) the word pokok
(‘main’/‘fundamental’/‘basic’) is suffixed with -nya, indicating the speaker’s atti-
tude toward the rest of the clause as being central to the understanding of the
ongoing interaction. Here, in IU 1248 of example (11), the speaker is highlighting
the fact that the women looked rich, as essential background for understanding
why they were being targeted by pickpockets. The corpus contains 53 similar in-
stances of pokoknya (‘main’-nya), suggesting that this is a highly grammaticized
way of highlighting the main point of the ongoing interaction. Following are two
more examples of -nya constructions used as a stance marker of interactional
relevance.
(12) Pencuri IU 994

inti-nya dia malam itu nggak bisa n-emu-i
gist-stm 3sg night that neg can at-meet-app
saudara-nya itu.
sibling-pos that
The gist of it was that he couldn’t meet his brother that night.
In this example, the speaker is concluding a narrative told to her by another per-
son. She uses the noun inti ‘gist/nucleus/core’ suffixed with ‑nya to indicate her
evaluation of this utterance as being the gist or summary of the surrounding nar-
rative. As in the previous example, the speaker uses the -nya construction here
as an assessment of interactional relevance – indicating the degree of value of
the current utterance regarding its contribution to the ongoing discourse. Other
similar expressions found in the corpus include misalnya ‘for example’ (‘example’
-nya), used by a speaker to frame the utterance as a specific, usually hypothetical
example of the overall general point of the discourse; masalahnya ‘the problem
is’ (‘problem’-nya) which indicates that the current utterance is somehow prob-
lematic in light of the ongoing interaction; as well as pokoknya ‘the thing is’ and
intinya ‘the gist of it is,’ which have been illustrated in the previous examples.
The third type of stance indicated by -nya constructions relates to affect – the
speaker’s emotional attitude toward the rest of the proposition in the utterance.
The following three examples are illustrative.
(13) Pencuri IU 545–546

545 A: Takut-nya tu,
fear-stm that
546 mereka bawa benda tajam,
3pl bring object sharp
(We’re) afraid they might bring sharp objects.
In this example, the verb takut ‘fear’ is suffixed with -nya, indicating the speaker’s
emotional attitude toward the proposition in the rest of the utterance. She is afraid
that the pickpockets might ‘bring sharp objects’ with them on the bus, and would
hurt anyone who tries to thwart their criminal activities. Takutnya (‘fear’-nya) in
the first IU thus frames the next IU in terms of the speaker’s emotional attitude.
The following example is similar.
(14) Blewah IU 13
... Takut-nya lupa ya?
fear-stm forget prt
Are you afraid you’ll forget?
As in the previous example, the verb takut ‘fear’ is suffixed with -nya, framing the
proposition in the rest of the clause ‘(you’ll) forget’ as being something to poten-
tially fear. The following example illustrates another type of mental/emotional
attitude, namely the speaker evaluating the rest of the utterance as being lucky or
favorable.
(15) Pencuri IU 697

L: Untung-nya Si Agnes ke-rasa ya?
fortune-stm prt Agnes nonvol-feel prt
It was fortunate that Agnes felt it, right?
This clause serves as the coda to a story about one of the speakers and her friend
Agnes, who thwarted an attempted robbery on a bus because she felt someone
trying to undo her necklace. Speaker L uses the noun untung ‘fortune/luck’ with
the -nya suffix to indicate her emotional attitude toward the rest of the clause as
being favorable; it was ‘fortunate’ that ‘Agnes felt it.’ In sum, I have shown here
three examples of -nya being used as part of a construction to indicate affect – the
speaker’s emotional attitude toward the rest of the proposition. Examples (13)
and (14) indicate negative affect (specifically ‘fear’), while (15) illustrates positive
affect (the occurrence of an event being evaluated as ‘fortunate’).
To conclude, the current section has offered an overview of the ways in which
the -nya clitic is implicated in epistemic stance in colloquial Indonesian. Return-
ing to the definition which began this section, this type of stancetaking refers to “a
speaker’s indication of how he or she knows about, is commenting on, or is taking
an affective or other position toward the person or matter being addressed” (Wu
2004: 3). The three categories made explicit in Wu’s definition align closely with
the three main stance functions outlined here for -nya constructions. First, as an
evidential marker, -nya constructions indicate the speaker’s source of knowledge
for, or “how the speaker knows about” the utterance. Second, when marking the
interactional relevance of an utterance, -nya constructions serve to indicate how
the speaker “is commenting on” the utterance – assessing its value, usually to-
ward its contribution to the overall talk at hand. Thirdly, -nya constructions may
be used when the speaker “is taking an affective or other position toward” the
proposition in the rest of the clause, in which case they may indicate the speaker’s
emotional or mental attitude toward the event, e.g., as negative, as favorable, or
as something to be feared. I have presented numerous examples in this section
to illustrate the scope and variety of stancetaking in which the -nya clitic par-
ticipates. Yet, despite its pervasiveness in everyday language use, with the excep-
tion of Englebretson (2003) the stance-related functions of this clitic have been
virtually ignored in previous literature. Descriptive and pedagogical grammars of
Indonesian have tended to focus on the more general referential and grammatical
functions of -nya, such as third-person possession, definiteness, and nominaliza-
tion. In this section, I have shown that what has previously been described as
primarily referential and grammatical in nature is actually doing substantial work
in stancetaking. These findings offer another step toward the incorporation and
recognition of the centrality of stance in a description of Indonesian grammar.
5. Voice as a means of positioning
As a third illustration of general facets of Indonesian grammar that are used in

stancetaking, I conclude this paper with a discussion of the Indonesian grammat-
ical voice system of agent-trigger and patient-trigger clauses. I will demonstrate
that in addition to the syntactic, semantic, pragmatic, and discourse-structur-
ing features attributed to these constructions in previous literature, grammatical

voice may also serve as a resource for speakers to position themselves and others
(cf. Harré and van Langenhove 1999) in a moral landscape of intentionality and
responsibility. Self-expressive stancetaking was discussed in Section 3, epistemic
stancetaking was discussed in Section 4, and the type of stancetaking at issue in
the present section concerns the stances that speakers take about moral agency
and responsibility – either of themselves or of the people whom they are talking
about.
As with many Western Austronesian languages, the verb morphology of In-
donesian offers an array of voice and valence inflectional prefixes. Verbs may be
marked by the agent-trigger prefix meN-, (where N represents a nasal consonant
whose pronunciation is determined by the place of articulation and voicing of
the initial phoneme of the stem), which is often rendered as N- in colloquial In-
donesian; the patient-trigger prefix di-; other valence-marking prefixes such as
ber- (often glossed as a reflexive or middle-voice), ter- (whose meaning includes
nonvolitional, adversative, abilitative, and others), and ke- (also often glossed as
nonvolitional); or, as is commonly the case in colloquial Indonesian, no prefix at
all. The two prefixes I will focus on here are the agent-trigger meN-/N- and the
patient-trigger di-. Considerable debate exists in the Indonesianist literature as to
whether these prefixes represent active and passive respectively, antipassive and
ergative respectively, or something else (cf. Cumming and Wouk 1987; inter alia).
This debate is not relevant to the discussion here, so I have adopted the more
neutral terms agent-trigger and patient-trigger (cf. Cumming 1991; Englebretson
2003; Wouk 1989; inter alia). The following two examples give clause level illus-
trations of these prefixes as found in the corpus.
(16) Pencuri IU 753

Lagi mem-perbaik-i sepeda motor kan,
prog at-improve-app bike motor prt
(He) was fixing his motorcycle.
The verb in this example contains the agent-trigger prefix mem-, indicating that
the agent-argument of the verb memperbaiki ‘repair’ (in this case an unexpressed
3SG reference clear from the discourse context) is the subject of the clause, and
the patient-argument sepeda motor ‘motorcycle’ is the object.9 As suggested by
the free translation, Indonesian agent-trigger clauses often correspond to English
active voice. Following is an example of a patient-trigger clause.
(17) Pencuri IU 1263

Aku di-tahan sama temen-ku samping-ku.
1sg pt-restrain with friend-1sg side-1sg
My friend next to me held me back.
(lit., I was held back by my friend next to me.)
In this example, the verb tahan ‘restrain/hold back’ is prefixed with the patient-
trigger di-, indicating that the patient-argument aku (one of the 1SG pronouns
discussed in Section 3) is the subject of the clause. The agent-argument temenku
‘my friend’ is preceded by the preposition sama ‘with/by,’ which marks it as an
oblique.10 While it is true that Indonesian patient-trigger clauses appear structur-
ally similar to English passives, note that there are substantial functional differ-
ences between how each of these clause types are used in each language; therefore
a direct English translation of patient-trigger clauses can be awkward if not im-
possible, and is often not the most accurate choice. I have included a literal Eng-
lish translation in many examples, in addition to the idiomatic free translation, in
order to remind the non-Indonesianist reader of the structure of the Indonesian
clauses under discussion.
Substantial literature exists on the functions of agent- and patient-trigger
clauses in Indonesian. (See Kaswanti Purwo 1988 and Wouk 1989 for broadly
functionalist perspectives and overviews of relevant research.) General explana-
tions for their use tend to focus on syntactic factors (e.g., only the trigger ar-
gument may participate in relativization and act as the pivot in certain types of
clause combination); clusters of related semantic factors such as animacy, refer-
entiality, and individuation of the arguments, and the event structure and implied
aspect of the verb; and constellations of discourse and pragmatic factors such as
topicality, foregrounding and backgrounding, discourse transitivity, and narrative
versus non-narrative clause types. In the current section, I claim that an impor-
tant aspect of Indonesian voice has so far been overlooked. Namely, in addition
to previous explanations such as those just mentioned, the Indonesian voice sys-
tem has profound social implications for constructing moral agency, and for how
speakers position themselves and others. Following Duranti,
Agency is here understood as the property of those entities (i) that have some
degree of control over their own behavior, (ii) whose actions in the world affect
other entities’ (and sometimes their own), and (iii) whose actions are the object
of evaluation (e.g., in terms of their responsibility for a given outcome).
(2004: 453)
This tripartite approach to human agency (intentional control of one’s actions,

ability to affect others, and moral accountability) is fully relevant with regard
to the Indonesian voice system. The ready availability of agent-trigger and pa-
tient-trigger morphology facilitates the positioning of individuals as intentional
or non-intentional, as agentive or patientive, and as morally responsible and ac-
countable, or as morally not responsible and unaccountable. When viewed from
this perspective, the Indonesian voice system is far more than a means of express-
ing semantic meaning and discourse structure, as it takes on a central role in the
social worlds which speakers are constructing through talk.
In support of this claim, I offer one particularly rich stretch of talk from the
corpus, a conversational narrative which is overtly concerned with aspects of
agency and responsibility and the lack thereof. I will show how a speaker uses
agent- and patient-trigger clauses in stancetaking, throughout this excerpt, to po-
sition a pair of friends in terms of moral agency and responsibility. Because of the
lengthy nature of this excerpt, I cannot reproduce it here in its entirety. Rather, I
will summarize the excerpt, and then discuss the clauses which contain either an
agent-trigger or patient-trigger verb form.
The excerpt begins approximately 38 minutes and 15 seconds into the Pencuri
segment. Earlier, one of the speakers had commented that she had recently read
in the newspaper about thieves at the Beringharjo market (the large, traditional
Javanese market in Yogyakarta) who have been victimizing people by means of
Gendam. Gendam refers to a reputed form of black magic that combines hypnosis
along with mystical and supernatural elements. Practitioners of Gendam allegedly
can use their powers to victimize unsuspecting people in public places, by causing
them to temporarily lose full consciousness of their surroundings (e.g., the victim
gives valuables to a stranger, believing that person to be a family member), to mis-
take a worthless object for one of value (e.g., the victim buys what s/he believes to
be a watch, and discovers only after returning home that it is actually a rock), or
to believe that a poorly-made item is of exceptional quality (i.e., the victim spends
a large sum of money for something that turns out to be worth far less). One of
the important details about Gendam is that its influence only operates within a
small area around the practitioner. As a consequence, if one is shopping at the
market with a friend and notices that the friend has fallen victim to Gendam, the
unaffected person must get the victim outside of the boundaries of the Gendam
in order to break its power. The relevant excerpt for the current section consists of
speaker I (Indra) recounting an event that happened to two friends of hers named
Maya and Erika. I will briefly summarize Indra’s narrative. Maya and Erika are
shopping at the market when Maya comes under the influence of Gendam. Maya
sees some towels for sale, believes them to be big, thick, and of high quality, and
is getting ready to buy the towels. Erika notices that the towels are actually junk,
and realizes Maya is being influenced by Gendam. A struggle ensues between the
two women, as Erika tries to pull Maya away from the towels and out of the reach
of the Gendam. Eventually, Erika succeeds, and Maya comes far enough away
that she is no longer under its power. At this point Maya becomes fully conscious,
looks again at the towels, and realizes that they are not at all good or worth the
high price.
For purposes of the present discussion, what is especially interesting about
Indra’s narrative is her use of agent- and patient-trigger clause types. While Maya
is under the power of Gendam, Indra uses only patient-trigger clauses to refer to
her. In contrast, the only agent-trigger clauses found in this part of the narrative
are those with Erika as the agentive subject. After Maya has become fully con-
scious and is no longer affected by Gendam, she once again becomes the subject
of agent-trigger clauses too. In other words, in this narrative, Maya’s being under
the influence of Gendam correlates with an avoidance of placing her as an agen-
tive subject, but this restriction in grammatical roles is lifted once she is again ful-
ly conscious. I would suggest that the distribution of clause types in this narrative
is no accident. Rather, Indra is actively constructing a social world in which she
is using patient-trigger clauses to position Maya as a victim of Gendam, who is
thus not in control, not able to affect others, and not morally accountable for her
actions. At the same time, Indra is actively using agent-trigger clauses to position
Erika (the woman who was not influenced by Gendam) as intentional, agentive,
and responsible for removing Maya from the influence. Following is a discussion
of the specific relevant clauses from Indra’s narrative.
After Maya has been introduced into the narrative as Indra’s friend and a fellow
student of English, Indra summarizes Maya’s appraisal of the towels at the market.
These first seven IUs have already appeared above in example (5) (regarding the
use of the informal aku 1SG pronoun in constructed dialogue), and are repeated
here for convenience as example (18). These IUs serve as a general summary of the
initiating events and an orientation to Maya’s predicament at the market:
(18) Pencuri IU 2751–2757

2751 I: .. Maya itu udah ter-tarik banget.
Maya that perfv nonvol-pull very
I: Maya got really attracted.
2752 Ah murah ini,
oh cheap this
“Oh, this is cheap!
2753 aku pingin handuk tiga ribu itu.
1sg want towel three thousand that
I want this towel for 3,000 (Rupiah).
2754 Padahal gede tebal itu,

in:fact big thick that
It’s big and thick.
2755 harus‑nya enam ribu lebih to.
should‑stm six thousand more prt
It should be 6,000 (Rupiah) more.”
2756 dia kan,
3sg prt
She was (like),
2757 .. mikir‑nya ah lumayan @mahasiswa @baru @gitu ya.
thought‑pos prt reasonable student new thus prt
Her thoughts were reasonable for a new student, you know.
In IU 2751, Indra describes Maya as having been “very attracted” by something.

She uses the nonvolitional ter- prefix here. Maya is not presented here as an active
agent, but neither is she being construed yet as fully patientive. IUs 2752–2755
present Maya’s inner dialogue, and also the object of her interest – the towels that
she thinks are cheap, big and thick, and worth the price. In IUs 2756–2757, Indra
assesses Maya’s thoughts as being “reasonable for a new student” – presumably
because as a new student, Maya is likely short on money and in search of a good
bargain. Indra’s first introduction of Erika into the narrative follows:
(19) Pencuri IU 2760–2761

2760 .. Terus untung-nya Erika-nya itu,
next fortune-stm Erika-def that
2761 udah ng-erti itu lho.
perfv at-understand that prt
And then fortunately Erika understood (what was going on).
Here Indra is introducing Erika into the narrative. Significantly, she does so by
means of the verb mengerti ‘understand.’ This verb is relevant for two reasons.
First, it has the agent-trigger prefix, thereby construing Erika as an agent. Secondly,
the meaning of the verb itself implies that Erika is cognizant and aware of Maya’s
situation. The next segment of the narrative presents the struggle between the two
women, as the Gendam is pulling Maya in, while Erika is trying to pull her away.
(20) Pencuri IU 2764–2771

2764 di-tarik-tarik,
pt-pull-pull
(Maya) got pulled in and pulled in.
2765 (H) itu udah ng-otot banget.
that perfv at-persist very
But (Erika) kept really persisting.
2766 Si Maya itu kamu itu di-singkir-singkir kee-,11
prt Maya that 2sg that pt-pull:away-pull:away trunc
Kept pulling Maya away. (lit., Maya was pulled away.)
2767 Sampai ber-antem sendiri ber-dua itu.
until mid-quarrel self mid-two that
Until the two of them started fighting with each other.
2768 (H) Tapi Erika-nya itu ng-otot,
but Erika-def that at-persist
But Erika kept on persisting.
2769 .. kalau di-tarik,
if pt-pull
2770 .. pergi jauh
go far
2771 berapa meter dari situ,
how:many meter from there
She was pulled, far away, a few meters from there.
This brief excerpt contains three instances of the di- patient-trigger prefix, and
two instances of the N- agent-trigger prefix. All three of the patient-trigger pre-
fixes (IUs 2764, 2766, and 2769) occur on the verb tarik ‘pull,’ and all take the
patientive Maya as the subject. (The actual NP is ellipted in two of these instances,
but it is clear in the overall context that Maya is the subject argument.) Both of
the agent-trigger prefixes (IUs 2765 and 2768) occur on the verb otot ‘persist,’
indicating it is Erika, the subject of these clauses, who is being persistent in co-
ercing Maya away from the towels and away from the influence of Gendam. The
following example contains one more instance of Maya as the subject in a patient-
trigger clause, as well as speaker L’s co-construction of the upshot of the narrative,
namely that Erika had to take her outside of the boundaries of the Gendam.
(21) Pencuri IU 2779–2782

2779 I: .. Udah di-ta= rik sampai marah-marah
perfv pt-pull until angry-angry
Maya-nya itu.
Maya-def that
I: (Maya) was pulled until Maya started getting angry.
2780 L: [Pokok-nya di luar batas,
main-stm at outside limit
L: [The important thing is, outside of the boundary.
2781 I: [Sampai itu ho-o.
until that aff
I: [To there, yeah.
2782 .. di luar batas tu],
at outside limit prt
Outside the boundary].
Again, Maya is presented in a patient-trigger clause in IU 2779 as being pulled. In

the following example, Maya is once again presented as the subject argument in a
patient-trigger clause, as Erika is addressing her:
(22) Pencuri IU 2792–2795

2792 .. Langsung kamu ini gini-gini,
direct 2sg this like:this-like:this
“You just keep going like this.”
2793 pokok-nya di-omong-in sama Erika itu,
main-stm pt-talk-app by Erika that
Erika said to her. (lit., (she) was told by Erika.)
2794 Terus udah lewat batas berapa meter itu,
next perfv pass limit how:many meter that
Then, after they had passed the boundary of a few meters,
2795 .. Jadi sadari itu lho Maya-nya
become conscious that prt Maya-def
Maya became conscious.
In IU 2792, Indra is reporting the words that Erika allegedly said to Maya, en-
couraging her to keep coming with her. Because of the di- patient-trigger prefix
on the verb omong ‘speak’ and the oblique agent phrase sama Erika ‘by Erika,’ this
clause again construes Maya as the subject argument in a patient-trigger clause.
In IU 2794, Indra reports that they passed the boundaries of the Gendam, and IU
2795 provides the resolution of the narrative – when Maya jadi sadar ‘becomes
conscious.’ This marks a shift in positioning, as Maya is no longer portrayed as
the victim of Gendam, and this shift in positioning is also indicated by a shift in
grammar, as Indra begins to construe Maya as the subject in agent-trigger clauses,
as shown in the following example.
(23) Pencuri IU 2796–2799

2796 (H) O ya ya,
oh yes yes
“Oh, yeah yeah.”
2797 ternyata jelek nge-lihat lagi to,
in:reality bad at-see again prt
In reality it was bad when she looked at it again.
2798 n-engok lagi ke itu-nya.
at-look again to that-def
She looked at it again,
2799 .. Ternyata handuk-nya itu jelek.
in:reality towel-def that bad
In reality the towel was bad.
The first IU of this example presents Maya’s reported speech: two acknowledgement
tokens to presumably indicate that she is now aware of her situation. Then, the verb
lihat ‘see’ in IU 2797, and the verb tengok ‘look’ in IU 2798, are both marked as
agent-trigger. This indicates that Maya is once again intentional and agentive, and
is responsible for the actions of seeing and looking. And when Maya ‘sees’ and
‘looks,’ she recognizes that the towels were “in reality” bad, and not in fact what she
had thought they were when she was under the influence of Gendam.
In sum, the distribution of agent- and patient-trigger clauses in the unfold-
ing of this conversational narrative illustrates how speakers may use the general
grammatical resource of voice in Indonesian to position themselves and others
in terms of agency and responsibility. When Maya is under the influence of Gen-
dam, Indra (the narrator) positions her as patientive and not responsible for her
actions. Indra achieves this positioning, in part, through the use of patient-trigger
clauses for Maya, in contrast to the agent-trigger clauses she uses with Erika, who
is not under the influence of Gendam. Once Maya has come outside the bound-
aries of Gendam, Indra shifts her positioning through the use of agent-trigger
clauses, implying that Maya is once again able to control her actions and may now
be held accountable for them. Similar to Duranti’s (1990, 1994) analysis of erga-
tivity in Samoan public discourse, in which grammatical relations (specifically

ergative) allow speakers to assign blame or responsibility, the narrative analyzed
here suggests that Indonesian voice may have similar social implications.
The current section in no way seeks to discredit or minimize other, more tra-
ditional approaches to voice in Indonesian, which highlight semantic, syntactic,
or discourse factors. Conceiving of voice as a means of positioning the self or oth-
ers is not contradictory to approaching it from these other angles as well. How-
ever, placing central emphasis on the social nature of this phenomenon provides
a potentially fruitful line of inquiry into how speakers use grammar in stancetak-
ing, to actively construct their social worlds. As Duranti notes about agency,
a possible direction for future research is to expand our horizon of theoretical
and empirical research to include an understanding of these phenomena not only
from the point of view of the type of information that is being encoded (e.g. is the
agent of this event expressed and, if so, how? ) but also from the point of view of
the type of persons and the type of world that speakers build through their typi-
cally unconscious but nevertheless careful choice of words. (2004: 466)
In Indonesian, this “careful choice of words” must also be understood to encom-

pass the choice of agent- or patient-trigger clause, as a grammatical resource
which Indonesian speakers may use for specific stance purposes.
6. Conclusion
I began this paper by noting the pervasiveness of stancetaking as an activity real-

ized through language. I then tied this in with the functionalist view of a usage-
based approach to grammar, and suggested that one goal of the descriptive gram-
marian should be to account for how speakers use the grammatical resources of
a language to carry out the activity of stancetaking. Many English grammarians
and linguists have been engaged in such an enterprise over the course of the past
several decades, and it is not surprising, for instance, to find a chapter in the
Longman Grammar of Spoken and Written English (Biber et al. 1999) entitled The
grammatical marking of stance. However, interest in stancetaking and grammar
has not seemed to have had much influence on field linguistics or descriptive
grammars of other languages. As a first step toward such a goal, this paper has
offered a discussion of three facets of Indonesian grammar that speakers use in
constructing their social worlds through stancetaking.
In Section 3, I addressed the complex system of Indonesian first-person-sin-
gular referring expressions. I argued that these forms are not simply a determin-
istic reflection of a priori social variables such as regional origin of the speaker
or formality level of the speech event, but rather, speakers actively use these vari-
ous forms to construct and reflect locally-relevant aspects of social and personal
identity such as distance, casualness, or toughness. This kind of self-expressive
stancetaking enables speakers to highlight facets of the self, as relevant in the
ongoing conversation.
In Section 4, I took up the issue of the Indonesian -nya clitic and its partici-
pation in epistemic constructions. While this has traditionally been characterized
as a marker of possession, definiteness, or nominalization, I demonstrated that
much of its work in everyday interaction is to provide a means by which a speaker
can indicate epistemic stance – specifically toward the talk at hand. The epistemic
functions of -nya constructions fall into three categories: evidentiality (how a
speaker knows about the current utterance), assessment of interactional relevance
(how a speaker values the current utterance in light of the ongoing interaction),
and affect (the speaker’s emotional attitude toward the proposition expressed in
the current utterance).
Section 5 explored the social nature of Indonesian agent-trigger and patient-
trigger clauses. I argueed that grammatical voice, which has traditionally been
characterized solely in terms of syntax, semantics, and discourse-pragmatics, is
additionally a social resource for stancetaking that speakers can use in order to
position themselves and others in terms of moral agency and responsibility. I il-
lustrated this approach with reference to a specific conversational narrative, dem-
onstrating how the speaker uses grammatical voice to attribute agency, responsi-
bility, and the lack thereof to the protagonists in the narrative.
I have focused on these three aspects of Indonesian grammar for several rea-
sons. First, each is highly frequent and productive in colloquial Indonesian. Sec-
ond, these three grammatical resources are formally quite different from each
other: pronoun choice, a clitic, and voice constructions. Thirdly, each represents a
particular kind of stancetaking: self-expressive stance for the 1SG reference forms,
epistemic stance for the -nya constructions, and a broader-level stance in terms
of moral positioning for the agent-trigger and patient-trigger clause types. Yet,
these only begin to scratch the surface of how Indonesian grammar is involved
in stancetaking. I have not dealt at all with clause-final particles (cf. Wouk 1998,
2001), nor with modals and adverbs, nor with stancetaking through conversa-
tional sequences and dialogic interaction. What I offer here is a first glimpse of the
fruitful nature of consciously bringing stancetaking into the realm of grammati-
cal description. By having put these three areas on the table, I hope to encour-
age other linguists to take seriously the question of how grammatical resources
are functioning in the service of stancetaking in other languages. In this paper,
I hope to have shown that when we approach grammar with this perspective in
mind, forms which have previously only been characterized as part of a referen-
tial, propositional system for expressing and managing information can be seen
from a new perspective, a perspective that can offer fresh insight into the very
motivations behind grammar itself.
Notes
1. I would like to thank the participants in the 10th Biennial Rice Linguistics symposium for
comments and discussion on an early draft. The comments by the two anonymous reviewers
were likewise valuable, and helped to clarify several details. I would also like to thank Susanna
Cumming and Sandy Thompson, both of whom have provided invaluable encouragement and
suggestions at various stages of this project. As with all my work on Indonesian, I am especially
grateful to the individuals in Yogyakarta, who allowed themselves to be recorded for the col-
loquial Indonesian corpus, and especially to the three research assistants with whom I worked
through the data. All shortcomings of this paper, and any misunderstandings of Indonesian
language or culture are, of course, solely my fault and responsibility.
2. I wish to credit Jack Du Bois, in several unpublished conference talks, for presenting this
maxim: ‘Every utterance enacts a stance.’
3. I will not be discussing second- or third-person reference, nor the first-person-plural inclu-
sive and exclusive pronouns.
4. A reviewer has raised the provocative question as to the role of non-self-referral in the en-
actment of stance. This is indeed an important issue, which the author is beginning to address
in the context of a larger, cross-linguistic project on argument ellipsis, a discussion of which
is far afield of the current paper. In sum, considering this a colloquial Indonesian “zero 1SG
form” in opposition to the overt pronominal forms is problematic for at least two reasons. First,
this view presupposes that arguments of a clause are normatively expressed – which may not
in fact be true for colloquial Indonesian at all (cf. Ewing 2005). Secondly, argument ellipsis is
widespread in this language, and is by no means restricted to 1SG: it occurs with all persons of
pronouns, as well as with full noun-phrases. The use or ellipsis of arguments does indeed have
social and interactional implications – but whatever these implications are, they crosscut all
pronominal and full forms, and are not limited to self-reference or individual identity in the
same way as the different 1SG pronominal forms seem to be.
5. Due to the uncontrolled nature of the data, a direct statistical correlation for each of the
1SG reference forms across all six speech events is not feasible. Because the speakers vary con-
siderably among themselves in terms of the amount of talk, the topics, and their social back-
grounds, the confounding variables are too numerous to allow reliable conclusions to be drawn.
However, the categorical nature of the 1SG forms produced in the radio call-in show as com-
pared with the other five speech events is nonetheless striking.
6. There are actually two plausible structural analyses of this clause. In the one intended by
this example, duit gua is a possessive NP consisting of the head noun duit ‘money’ followed by
the 1SG form gua as a possessor: ‘my money.’ A literal English translation of this clause would
then be: ‘My money doesn’t exist.’ Another possible analysis of this clause would be as a type of
so-called focus construction, where nggak ada duit is the predicate, followed by the subject gua
‘I.’ According to this interpretation, the pronoun gua ‘I’ would be the subject of the clause, and
the overall focus construction has the pragmatic force of placing strong emphasis on the 1SG
subject “(As for me), I don’t have any money.” Both analyses are plausible, and, just as real-life
utterances are often uncertain and multifaceted for interlocutors themselves in realtime con-
versation, I see no need to choose one or the other structural interpretation here; they both ex-
ist here simultaneously. For purposes of exposition, I am presenting this example as an instance
of gua as the possessor in a possessive NP, although I recognize the other structural possibility
as well.
7. See Englebretson (2003: Chap. 5) for further discussion of the syntactic characteristics and
distribution of epistemic -nya constructions, and examples to support the analysis of these
constructions as adverbials.
8. This example shows -nya marking third-person-plural possession, which is generally con-
sidered a nonstandard use of this clitic. Prescriptively, -nya is said to mark third-person-sin-
gular possession. The extended use of -nya to mark possession in all persons and numbers is
another well-known difference between formal and colloquial varieties of Indonesian.
9. I am aware that the terms subject and object are debatable and problematic in the context of
Indonesian grammar. I am using these terms here solely for the sake of convenience, to facili-
tate the easy understanding of these examples by non-Indonesianist readers, and because the
debate over Indonesian grammatical relations is not central or relevant for the overall point of
this section.
10. In formal varieties of Indonesian, the preposition marking the oblique argument in a pa-
tient-trigger clause is oleh ‘by.’ The use of sama ‘with’ here is typical of colloquial Indonesian.
11. I am aware of the disfluencies and ambiguities in this IU, and the difficulty in glossing it in
a more accurate way. I will refrain from commenting further on this here, however, so as not to
distract from the discussion at hand.
References
Barlow, M. and Kemmer, S. (eds.). 1999. Usage-Based Models of Language. Stanford: CSLI.
Benwell, B. and Stokoe, E. 2006. Discourse and Identity. Edinburgh: Edinburgh University
Press.
Biber, D. and Finegan, E. 1988. “Adverbial stance types in English.” Discourse Processes 11 (1):
1–34.
of evidentiality and affect.” Text 9 (1): 93–124.
Biber, D., Johansson, S., Leech, G., Conrad, S. and Finegan, E. 1999. The Longman Grammar of
Brown, R. and Gilman, A. 1960. “The pronouns of power and solidarity.” In Style in Language,
T. Sebeok (ed.), 253–276. Cambridge, MA: MIT Press.
Discourse Studies 7 (4–5): 585–614.
Bybee, J.L. and Hopper, P.J. (eds.). 2001. Frequency and the Emergence of Linguistic Structure.
Bybee, J.L. 2006. “From usage to grammar: The mind’s response to repetition.” Language 82 (4):
711–733.
Cumming, S. 1991. Functional Change: The Case of Malay Constituent Order. New York: Mou-
ton de Gruyter.
Cumming, S. and Ono, T. 1997. “Discourse and grammar.” In Discourse Studies: A Multidisci-
plinary Introduction, Vol. 1, T.A. van Dijk (ed.), 112–137. London: Sage.
Cumming, S. and Wouk, F. 1987. “Is there ‘discourse ergativity’ in Austronesian languages?” In
Studies in Ergativity, R.M.W. Dixon (ed.), 271–297. Amsterdam: North-Holland.
Du Bois, J.W., Schuetze-Coburn, S., Paolino, D. and Cumming, S. 1993. “Outline of discourse
transcription.” In Talking Data: Transcription and Coding Methods for Language Research,
J.A. Edwards and M.D. Lampert (eds.), 45–89. Hillsdale, NJ: Lawrence Erlbaum.
Duranti, A. 1990. “Politics and grammar: Agency in Samoan political discourse.” American
Ethnologist 17 (4): 646–666.
Duranti, A. 1994. From Grammar to Politics: Linguistic Anthropology in a Western Samoan Vil-
lage. Berkeley: University of California Press.
Duranti, A. 2004. “Agency in language.” In A Companion to Linguistic Anthropology, A. Duranti
(ed.), 451–473. Malden, MA: Blackwell.
Echols, J.M. and Shadily, H. 1989. An Indonesian-English Dictionary, 3rd ed. (revised and edited
by John U. Wolff and James T. Collins in cooperation with Hassan Shadily). Ithaca: Cornell
University Press.
Englebretson, R. 2003. Searching for Structure: The Problem of Complementation in Colloquial
Indonesian Conversation. Amsterdam: John Benjamins.
Errington, J.J. 1988. Structure and Style in Javanese: A Semiotic View of Linguistic Etiquette.
Philadelphia: University of Pennsylvania Press.
Errington, J.J. 1998. Shifting Languages: Interaction and Identity in Javanese Indonesia. New
York: Cambridge University Press.
Ewing, M.C. 2005. “Colloquial Indonesian.” In The Austronesian Languages of Asia and Mada-
gascar, A. Adelaar and N. Himmelmann (eds.), 227–258. London: Routledge.
Ford, C.E., Fox, B.A. and Thompson, S.A. 2002. “Social interaction and grammar.” In The New
Psychology of Language. Vol. 2, M. Tomasello (ed.), 119–143. Mahwah, NJ: Lawrence Erl-
baum.
Givón, T. 1979. On Understanding Grammar. New York: Academic Press.
Harré, R. and van Langenhove, L. 1999. Positioning Theory: Moral Contexts of Intentional Ac-
tion. Malden, MA: Blackwell.
Hopper, P.J. 1987. “Emergent grammar.” BLS 13: 139–157.
Hunston, S. and Sinclair, J. 2000. “A local grammar of evaluation.” In Evaluation in Text: Au-
thorial Stance and The Construction of Discourse, S. Hunston and G. Thompson (eds.),
74–101. Oxford: Oxford University Press.
Kamio, A. 1997. Territory of Information. Amsterdam: John Benjamins.
Kaswanti Purwo, B. 1988. “Voice in Indonesian: A discourse study.” In Passive and Voice, M.
Shibatani (ed.), 195–241. Amsterdam: John Benjamins.
Langacker, R.W. 1999. Grammar and Conceptualization. New York: Mouton de Gruyter.
Mühlhäusler, P. and Harré, R. 1990. Pronouns and People: The Linguistic Construction of Social
and Personal Identity. Cambridge, MA: Blackwell.
Ochs, E., Schegloff, E.A. and Thompson, S.A. (eds.). 1996. Interaction and Grammar. New York:
Cambridge University Press.
Ono, T. and Thompson, S.A. 1995. “What can conversation tell us about syntax?” In Alterna-
tive Linguistics: Descriptive and Theoretical Modes, P.W. Davis (ed.), 213–271. Amsterdam:
John Benjamins.
Poedjosoedarmo, S. 1982. Javanese Influence on Indonesian. Materials in Languages of Indone-
sia, No. 7. Series D. Canberra: Pacific Linguistics.
Rampton, B. 2005. Crossing: Language and Ethnicity among Adolescents, 2nd ed. Northampton,
MA: St. Jerome Pub.
Sneddon, J.N. 1996. Indonesian: A Comprehensive Grammar. New York: Routledge.
Sneddon, J.N. 2002. “Variation in informal Jakartan Indonesian: A quantitative study.” Paper
presented at the Ninth International Conference on Austronesian Linguistics, Canberra,
January 8–11.
Stubbs, M. 1986. “‘A matter of prolonged field work’: Notes towards a modal grammar of Eng-
lish.” Applied Linguistics 7 (1): 1–25.
Tannen, D. 1986. “Introducing constructed dialogue in Greek and American conversational
and literary narrative.” In Direct and Indirect Speech, F. Coulmas (ed.), 311–332. New York:
Mouton de Gruyter.
Tannen, D. 1989. Talking Voices: Repetition, Dialogue, and Imagery in Conversational Discourse.
New York: Cambridge University Press.
Thompson, G. and Hunston, S. 2000. “Evaluation: An introduction.” In Evaluation in Text: Au-
thorial Stance and the Construction of Discourse, S. Hunston and G. Thompson (eds), 1–27.
Oxford: Oxford University Press.
Tomasello, M. (ed.). 1998. The New Psychology of Language: Cognitive and Functional Ap-
proaches to Language Structure, Vol. 1. Mahwa, NJ: Lawrence Erlbaum.
Tomasello, M. (ed.). 2002. The New Psychology of Language: Cognitive and Functional Ap-
proaches to Language Structure, Vol. 2. Mahwa, NJ: Lawrence Erlbaum.
Wouk, F. 1989. The Impact of Discourse on Grammar: Verb Morphology in Spoken Jakarta In-
donesian. Ph.D. dissertation, Department of Linguistics, University of California, Los An-
geles.
Wouk, F. 1998. “Solidarity in Indonesian conversation: The discourse marker kan.” Multilingua
17 (4): 379–406.
Wouk, F. 1999. “Dialect contact and koineization in Jakarta, Indonesia.” Language Sciences 21
(1): 61–86.
Wouk, F. 2001. “Solidarity in Indonesian conversation: The discourse marker ya.” Journal of
Pragmatics 33 (2): 171–191.
Wu, R.-J. 2004. Stance in Talk: A Conversation Analysis of Mandarin Final Particles. Amsterdam:
John Benjamins.
Appendix A: Summary of transcription conventions
(Adapted from Du Bois et al. 1993)

Each transcript line represents a single intonation unit.
Speaker labels appear in uppercase, and are followed by a colon.
Simultaneous speech is indicated by square brackets [ ] (not aligned because of glossing dif-
ficulty).
Material in angle-brackets < > is codeswitching, usually Javanese, English, or Arabic.
. Final intonation contour

, Continuing intonation contour
? Appeal intonation contour
‑‑ Truncated intonation unit
‑ Truncated word
@ One pulse of laughter
% Glottal stop
= Prosodic lengthening
.. Short pause
... Long pause
(TSK) Tongue click
(H) In-breath
Appendix B: Abbreviations and glosses
1sg, 2sg, 3sg first-, second-, and third-person singular

1pl, 3pl first-person plural, third-person plural
aff affirmative backchannel
app applicative suffix
at agent-trigger prefix
def definite
ex existential
mid middle voice
neg negative morpheme
nom nominalizer
nonvol nonvolitional
perfv perfective aspect
pos possessive
prog progressive aspect
prt discourse particle
pt patient-trigger
rel relativizer
stm stance marker (epistemic -nya)
Subjective and intersubjective uses
of generalizations in English conversations
Joanne Scheibman
Old Dominion University
1. Introduction1
1.1 Introduction to the study
Generality of meaning has been of interest to scholars working primarily in the

areas of semantics, language acquisition, and language change. To date, however,
generalization and generality of meaning in discourse have not received a great
deal of attention, even though there is evidence that speakers’ uses of general
meanings have discursive and pragmatic functions. For example, included with
orientation and attitude, Berman (2005) lists generality of reference as one of three
dimensions of discourse stance. Van Leeuwan (1995) suggests that generalization
and underspecification of agency and social actions are important considerations
in critical analyses of texts. And Zhang (1998) points out that general expres-
sions rely on Grice’s maxims for their resolvability. These findings suggest that
generality of expression also has influences on social and expressive activities in
conversational discourse.
The present study examines stance-related functions of generalizations in
English conversations. These are expressions of position and attitude that are rel-
evant both to individual speakers (their subjective uses) and to relational activities
among participants (their intersubjective uses). These utterances index and evalu-
ate classes in American English conversations. The expressions designate types –
as opposed to tokens – of people, things, attitudes, reactions, activities, and rela-
tions. Additionally, in the conversations used for this study, generalizations based
on these classes are not inductively constructed or construed – that is, they are
not categories based on reported properties shared by multiple entities or ideas
in the world. Instead participants tend to generalize (index classes of experience)
112 Joanne Scheibman
based on personal and social expectations and beliefs, and the use and interpreta-
tion of these utterances is keenly sensitive to interactive contexts.
The specific types of generalizations examined here are utterances with gram-
matical subjects that refer to classes or groups (e.g., and all these teachers are com-
ing in and saying; worse things could happen to the poor guy; the French are basically
a northern people). Because construction and interpretation of meaning and refer-
ence in conversation occur interactively, participants’ classifications of people and
phenomena for purposes of predication have expressive and social consequences.
As analyses offered in this paper suggest, generalizations in these conversations
are used by participants to evaluate, demonstrate solidarity with one another, and
authorize opinions – activities which, in many cases, animate societal discourses.
Moreover, because these utterances are naturally broadening, or inclusive, in the
sense that they typically index commonly held beliefs, their uses in conversations
highlight collaborative and interactive aspects of stancetaking.
1.2 Generality, generalizations, and stance
In a description of a crosslinguistic project investigating the development of stance

over time based on spoken and written narratives, Berman (2005: 107) character-
izes discourse stance as having three related dimensions: orientation (relations be-
tween sender, text, and recipient), attitude (epistemic, deontic, affective), and gen-
erality (of reference and quantification). The category generality “concerns how
relatively general or specific reference is to people, places, and times mentioned in
the text,” and in these studies textual features are coded according to three values:
personal and specific (e.g., I/my parents think, My/This boy’s father made me/him
apologize), generic (e.g., People/We/You tend to think), and impersonal (e.g., It’s
well known) (Berman 2005: 108). Berman operationalizes referential generality,
then, as a continuum – with linguistic features at one end categorized as personal
and intimate, and those at the other end viewed as impersonal and distant.
The set of utterances defined as generalizations in the current study includes
those in Berman’s generic group and also contains expressions with lexical noun-
phrase subjects that refer to general classes. The functions and properties of these
generalizations also overlap with Berman’s orientation and attitude dimensions
of discourse stance. For example, even though utterances exhibiting the formal
properties of personal and specific expressions as described by Berman are not
included in this study (e.g., those with specific agent subjects such as I), the con-
versational generalizations are “personalized” in a different sense: they convey
subjective meanings (primarily evaluation) in interactional contexts. Indeed, one
Generalizations in English conversations 113
of the consistent findings of the analyses presented here is that in interactive con-
texts, the generalizations function evaluatively in their conversational contexts.
Stance in discourse is a relational notion. Stancetaking is a reaction to, or a po-
sitioning relative to, something else – prototypically, a message or topic. Berman
suggests that stance “reflects a key facet of human discourse in general: the fact
that any state of affairs in the worlds of fact or fantasy can be described in multiple
ways” (2005: 109). Speakers and groups of speakers, however, don’t only position
themselves with respect to propositional material, they also situate themselves in
relationship to one another, to expectation, and to sociocultural beliefs. While it
is not unusual for studies of stance to focus on the use of linguistic features by
individual speakers and writers to represent events and entities in the world (e.g.,
Biber and Finegan 1989; Berman 2005; Reilly et al. 2005), this paper is informed
by the understanding that speakers also express stance by allying themselves with
(or sometimes separating themselves from) one another. In these ways, stances
can be collaboratively constructed and jointly held. One might argue that all dis-
cursive and social practices reflect stancetaking, since these activities presuppose
a response to some linguistic, psychological, or social reference point (Langacker
1993).2 Depending on what and whom interactants position themselves relative
to (e.g., utterances, objects, discourse participants, groups of people, beliefs), for
analytical purposes stance might be classified as individual (construed as relevant
to a speaker’s position in discourse), interactive (relevant to local discourse activi-
ties), or sociocultural (relevant to general beliefs of people as members of com-
munities). In the organization of this study, aspects of individual stancetaking
(evaluation, expressiveness) are framed as subjective, while interactive activities
and the sharing of beliefs contributing to collaborative expression of evaluation
and attitude are discussed as intersubjective. In the conversational extracts, how-
ever, the subjective and intersubjective functions of generalizations are typically
overlapping or simultaneous.
Generalizations have two related characteristics that contribute to their social
and expressive force in English conversations: (1) they make reference to (and as-
sess) general classes, (2) which gives them broadening, or inclusive, functions in
discourse. The predicates occurring with general subjects in these utterances are
often evaluative, a central type of subjective expression and stancetaking. It is also
the case that construction of an indexical class by a speaker is an expressive act.
Regarding the broadening properties of generalizations, because these utterances
are referentially general, to varying degrees they are underspecified. Their unre-
markable interpretation and use in conversations, however, suggest that partici-
pants elaborate the meanings of these expressions by making reference to shared
knowledge and to interactive contexts, aspects of intersubjectivity in discourse. In
these conversations, then, general assertions and evaluations are likely to be com-
mon ground among interactants (Clark 1996). Additionally, because of the inclu-
siveness of these utterances, generalizations can be used as expressions of solidar-
ity, an intersubjective activity that contributes to the collaborative construction of
stance in conversations.
1.3 An example of a conversational generalization
In the conversational extract in (1), the subject noun phrase the people I work with
delineates a group of people who are characterized by the speaker as sometimes
being really stupid. Since it is unlikely that the speaker has assessed the intelli-
gence of all of the employees of the medical laboratory where she works, evoking
this discursive class, and its interpretation by interactants, occur principally in
relation to the speaker’s expressed stance toward her coworkers.
(1) I don’t know,

the people I work with,
sometimes are really stupid,
and I--
and I think I’m starting,
to figure them out a little bit more,
The speaker’s generalizing her opinion to the entire group of people she works
with – as opposed to simply selecting the relevant individuals – is functional in
the conversation. In structuring conversational narratives, speakers may intro-
duce people, places, and situations in general terms in order to orient participants
to their stories (Labov and Waletzky 1967). It can also be the case that by expand-
ing a reference class subsumed by an evaluation using a generalization, a speaker
broadens the domain of her assertion, thus implicitly augmenting its evidential
weight or appeal in the interaction (see, in particular, example 8). Furthermore,
the evaluative stances expressed by speakers using these utterances with general
subjects tend to reflect meanings and expectations that are presupposed by oth-
er participants in the conversations. For example, in (1), the indexical category
denoted by the people I work with is a common one in mainstream American
English-speaking communities, and as such, it invites participants to fill in and
elaborate the meaning of this expression to include, for example, the familiar and
sometimes familial attitudes people hold toward this unique group of people they
see on a daily basis – an interpretation that is compatible with the speaker’s evalu-
ation of the group in the generalization in (1). It is not unusual, then, in these
conversations for a speaker’s stance toward a generally construed class (of people,
relations, etc.) to reflect evaluations that are shared by other discourse partici-
pants. This indexing of jointly held beliefs by these general utterances has conse-
quences for both interpretation (e.g., enriching of meanings of these expressions
for participants) and interaction (e.g., demonstrating solidarity by acknowledging
shared stance and attitude).
What is paradoxical about generalizations found in the conversational data
is that because these utterances make reference to general classes of phenom-
ena, there is a sense in which they can be viewed by interactants as proposition-
ally powerful. That is, utterances containing general subjects (e.g., the best things
are German) are often treated as matter-of-fact descriptions of situations in the
world. In spite of this appearance, generalizations in conversational usages are
necessarily selective in the sense that what speakers choose to generalize about
is contingent on expressive and interpersonal exigencies of both the local con-
text and larger culture. In conversational contexts, then, these utterances can be
used by interactants as expressions of stance and solidarity – to evaluate or to
show alliance with other participants by confirming adherence to particular so-
cietal discourses. These two characterizations of generalizing utterances – their
local, subjective construction and their intersubjective treatment as statements of
norms – inform the analyses offered below.
The next section surveys popular and scholarly uses of the term generaliza-
tion itself, leading to a working characterization of the concept to be applied to
generalizations in the conversational data. Section 3 describes and illustrates the
formal properties of generalizations in the database used for the study. Sections 4
and 5 present analyses of generalizations from a conversational corpus. Section 4
discusses generalizations that function as shared evaluations, and Section 5 illus-
trates two ways in which generalizations can augment speaker stance in interac-
tive discourse. And finally, Section 6 presents summaries and conclusions.
2. Generalizations
2.1 Popular and scholarly characterizations of generalization
The meaning of generalization varies by context and discipline. Operationalizing

the term based on its multiple uses has been one of the most challenging aspects
of this project. Furthermore, because in Western cultural discourses the mak-
ing of general statements about the world is an activity that can be elevated or
deprecated, generalization itself is subject to social valuation. In some cases, gen-
eralization is viewed negatively. In high school and university writing classes in
the United States, for example, students are routinely directed to provide specific
examples to support their generalizations, which might be defined as ‘unsubstan-
tiated opinions.’ In contemporary public discourse in the United States and else-
where, generalizations over the activities and attributes of (usually marginalized)
groups of people are also viewed negatively, and in some contexts, warrant the
label bigotry. On the other hand, positive and authoritative treatment of general-
izations is apparent in Western academic and scientific disciplines in which gen-
eralizations have an influential role in construction of theories (e.g., in statements
of principles, Kafura 1998: 26). The generalizations that have value in Western
cultures tend to be those that are overtly inductive, that is, those conclusions that
have explicit links to specific facts or cases. In contrast, general statements based
on implicit or insufficient data tend to be viewed unfavorably (e.g., stereotypes).
Processes of generalization figure prominently in linguistics research. In
functional and cognitive usage-based approaches to morphosyntax, for example,
analysts discuss technical aspects of the formation, change, and acquisition of
grammar as generalizations over specific usages. Ronald Langacker characterizes
grammatical patterns as “schematic symbolic units, which differ from other sym-
bolic structures not in kind, but only in degree of specificity” (1987: 58, bolding
in original). In studies of language change, individual grammaticizing elements
have been analyzed as undergoing processes of generalization of meaning (reduc-
tion in semantic specificity) accompanied by generalization (expansion) of the
item’s contexts of use (e.g., Bybee and Pagliuca 1985; Bybee et al. 1994). And in
language acquisition research, children are assumed to be learning generaliza-
tions (abstract patterns of form-meaning correspondences) when acquiring the
semantics and grammar of their first language, such as those contributing to the
system of argument structure in English (Goldberg et al. 2004).3 What unites the
meanings of the processes of generalization in these language studies is that in
each case there is a weakening of formal and semantic specificity of individual
linguistic expressions based on experiences of use, with a concomitant widening
of applicability (e.g., of context or process), which contributes to the development
or change of linguistic patterns.
Clearly the generalizations presented in this survey are of different types and
have distinct goals. The generalizations of scientific theorists are analytical con-
clusions based on particular types of data or texts. The generalizations illustrated
for linguistics, on the other hand, tend to be descriptions of cognitive processes
relevant to language use and change. What these different treatments of general-
ization share, however, is that they refer to the construction or highlighting of a
class (e.g., a theoretical principle, a grammatical process, an opinion) (Bouacha
1995), often accompanied by an analytical focus on the attenuation in specificity
or salience of individual elements that are in a relation to the class (e.g., weaken-
ing of lexical meaning of grammaticizing elements; the prominence and longevity
of scientific principles over time relative to the individual data analyses that sup-
port them).4
Generalization, then, is a type of categorization. In contrast to treatments of
categorization, which tend to highlight membership criteria and relationships
among category members, the process of generalization focuses on the forma-
tion of the class, or category, itself. Analogous to the way membership in classi-
cal categories has been delineated based on shared properties of members (see
Lakoff 1987, e.g., for a critique of this view), formal treatments of generalization,
such as those used in software and database design, define the term in a similar
way. For example, Kafura states that “[g]eneralization identifies commonalities
among a set of entities. The commonality may be of attributes, behavior, or both”
(1998: 26). However, as illustrated in this section, not all processes of generaliza-
tion designate classes that are based on shared properties of multiple entities (e.g.,
the maligned generalizations of the composition student). In academic argumen-
tation and theory-building, the authority of generalizations is explicitly tied to
common findings of individual studies and analyses. But in non-scientific regis-
ters and genres (e.g., conversational interaction), the common details that inform
the construction of speakers’ generalizations are often inexplicit and covert.
2.2 Generality of expression in English conversations
Expression of general meaning is ubiquitous in English conversations. While it is

the case that conversational interactants can (and do) engage with one another at
various levels of specificity, in previous analyses of English conversation, I found
that a relatively small percentage of tokens in a conversational corpus could be
described as referentially specific, despite the attention such expressions enjoy in
linguistic theories and practices (Hopper 1997; Scheibman 2002; Silverstein 1976).
These more specific utterances are most frequently expressions with third-person-
singular subjects with human referents occurring with past-tense material predi-
cates (e.g., and he did a blood test) or oblique predicates (e.g., my test was Friday).
They tend to cluster in descriptive narratives and in settings in which participants
are involved in activities in which objects are focal (e.g., opening birthday pres-
ents). More frequent in these data are utterances that convey general meanings.
Because generalizations are common linguistic practices in discourse, they
are socially situated. For example, variation in generality and specificity in the
representation of social groups and actions in discourse can be ideologically con-
sequential. Van Leeuwan suggests that generalization is “an important issue in
critical discourse analysis, as texts which are mainly concerned with legitimizing
or de-legitimizing actions and reactions tend to move high up on the generaliza-
tion scale, including only the names of episodes, or of whole social practices”
(1995: 99). With respect to generalizations in English conversations, the expe-
riences, beliefs, and attitudes that interactants generalize over in collaborative
contexts are both sensitive to and influence interaction. In contrast to formal or
academic generalizations, the classes indexed by conversational generalizations
might be said to be “empty” categories in the sense that membership is often im-
plicit and covert. The data presented here suggest that speakers use these general-
izations to evaluate and strengthen stance; and to create intersubjective ties both
by generalizing experience and attitude and by ratifying others’ points of view
while mutually adhering to societal discourses.
3. Data
3.1 Generality of meaning, generics, and generalizations
Van Leeuwen (1995) notes that determining the linguistic properties that con-
tribute to generality of expression can be challenging. Because texts typically do
not display multiple representations of the same actions and events at different
levels of specificity, systematic semantic comparison is difficult. Based on cross-
linguistic studies on the development of stance, Berman points out that the same
text “may be both specific and general in reference to persons, places, and times”
(2005: 109). Because of this formal and functional heterogeneity (illustrated be-
low for English), for this study it was necessary to select for analysis an utterance
type that consistently conveyed generality of meaning in the conversational data.
Generalizations, as delimited here, are expressions that lack specific reference
(e.g., to time, place, event) and the formal presence of a speaking subject – both of
these being properties of generics as well (Bouacha 1995). Lyons (1977: 193–194)
states that generic propositions are tenseless, timeless, and aspectless, whereas
general referring expressions can occur in sentences that express time-bound
propositions. Lyons (1977: 194) also notes that the status of “generic, as distinct
from general, reference” is philosophically controversial. Bouacha agrees that it is
not easy to distinguish generalizations from generics, suggesting that the clearest
difference is that strict generics (e.g., Beavers build dams) are primarily found in
scholarly metalanguage, whereas generalizations (e.g., Cauliflower does well here)
are quite common in naturally-occurring discourse (1995: 47). The analyses in
the present study do not take into account differences between general and ge-
neric expressions. Rather, all subject noun phrases in the database that designate
classes, including the so-called generic pronouns you and they, are classified as
general subjects (Section 3.3).
English speakers generalize about people, events, time, place, reactions, and
attitudes. Therefore, a variety of grammatical and lexical constructions can ex-
press generality of meaning. For example, in conversation, plural noun phrases
regularly denote classes (e.g., the prices are so much lower there than here; he
threatened little kids). Adverbials of time (e.g., so you go and you practice on the
weekends), manner (e.g., they come in with that attitude), and place (e.g., grow-
ing up around here you would know better) also index classes of expectation and
experience. General construal and expression of events can be expressed aspectu-
ally, for example, with progressive constructions (e.g., all the teachers were telling
me) or verbs in the habitual present (e.g., they play salsa?). Because of this struc-
tural and semantic heterogeneity of generalizing types in English, analyses in this
paper are restricted to conversational utterances with grammatical subjects that
designate classes (Section 3.3). As it turns out, expressions with general subjects
tend also to contain generalizing predicates (Section 3.4).
3.2 Sources of data
The database of utterances from which the generalizations were drawn consists
of portions of 10 conversations, totaling approximately 90 minutes. Five of the
conversations are from the Santa Barbara Corpus of Spoken American English
(Du Bois et al. 2000; Du Bois et al. 2003), and five are unpublished transcripts.5
There are 153 utterances with general subjects in the corpus representing the
language of 19 adult speakers of American English. Utterances were tagged for
a variety of structural and functional properties for a previous study (e.g., tense,
verb type, subject animacy, and referentiality) (Scheibman 2002). A coding field
was added specifying the general class indexed by the grammatical subject, as
illustrated in (2).
(2) the English are just the same though,6

(class: nationality)
your people .. tortured him to death,
(class: religious group)
everybody in the world has those problems
(class: hyperbolic subject)
3.3 General subjects
As stated above, the generalizations examined in this study are utterances with
grammatical subjects that designate classes. While it is the case that speakers evoke
classes in object or oblique positions (e.g., He knows Cats;7 I’m not dancing with
guys), English subjects have interactive and cognitive importance, which makes
utterances with general subjects a useful choice for these analyses. Chafe (1994)
suggests that English subjects are often starting points for additional information
(MacWhinney 1977), and Langacker (1993) points out that topics (and subjects)
can serve as reference points for further talk. In this study, classes indexed by
subject noun phrases are evaluated and elaborated in accompanying predicates,
resulting in full utterances that convey generality of meaning.
Table 1 summarizes the generalizations in the database by subject type. The
three major subjects that refer to general classes are generic they and you forms
and lexical noun phrases (plural and singular).8 The classes referred to by lexi-
cal noun-phrase subjects in the database (and anaphoric they subjects as well)
are diverse and vary in their conventionality. They include, for example, concrete
items and groups of people expressed with plural nouns (e.g., vacuum cleaner
bags, the women), institutional entities (e.g., the press), and hyperbolic subjects
(e.g., everybody in the world). Generic you utterances tend to universalize experi-
ence in conversations (e.g., you don’t admit you’re a feminist), and generic they is
often used in these conversations to characterize people or institutions construed
as outsiders (e.g., you can’t really tell when they blush).
Table 1. Generalizations by subject type (n = 153)

Subject type Examples
they 74 (48%) so they don’t know what the hell they’re doing
and in high school you know they teach about it
you 31 (20%) you can’t kill peas
growing up around here
you would know better
Lexical noun phrase 30 (20%) the ballroom people don’t do it that way
(plural) those guys’ll be all over you
Lexical noun phrase 16 (11%) the ceviche’s full of vegetables
(singular) the press will pick it up and smear you for it
one 2 (1%) one is more tolerant of feminist views
one’s certainly more sensitive to them
3.4 Predicates occurring with general subjects
The formal properties of predicates occurring in utterances with general subjects

also convey general meanings. Table 2 displays the generalizations by tense and
modality. Notably, 71 percent of these predicates (109/153) are present tense,
while only 12 percent are past tense. Additionally, 82 percent of the present-tense
utterances (89/109) are simple present, expressing stative and habitual meanings.
The examples in Table 2 illustrate how these predicates express processes and
states as non-specific.
Seventeen percent of the predicates occurring with general subjects (26/153)
contain a modal auxiliary. Half of these occur with generic you subjects, and the
most frequent modals in these utterances are can and can’t (69% or 9 out of 13
tokens). Many of these utterances are normative statements.
In contrast to the high frequency of present tense in these utterances, only 12
percent (18/153) of the predicates occurring with general subjects in the database
are past tense. This means that there are relatively few predicates in the group that
might designate referentially specific events (e.g., I talked to a guy that’s thirty-four
in my class,). As is the case, however, the past-tense predicates occurring with
general subjects tend to reflect construal of events as summaries, as opposed to
individual accounts. For example, in (3), line 1, Sharon uses a general subject and
a past-progressive predicate to present a series of similar events. In the utterance,
All the teachers were telling me, the speaker is not emphasizing individual tellings
of specific teachers at different times. Rather, Sharon’s use of all and a progressive
construction highlights her general reaction to her fellow teachers’ advising her
(that all of them were doing it), and not to the details of each encounter. This
analysis is supported by repetition of the teachers’ advice in the quoted material
in lines 3 and 4.
(3) (Raging Bureaucracy SBC0004 670.65-674.41)9

1 → SHARON: All the [teachers were] telling me,10
2 SHANE: [Could be worse].
3 SHARON: You’re too nice,
4 you’re too nice,
5 your kids are gonna take advantage of you
Table 2. Generalizations by tense and modality (n = 153)

Examples
Present 109 (71%) people have the information they need (stative)
everything breaks on my kitchen floor (habitual)
Modal 26 (17%) you can go up there
you can’t constantly be torn
Past 18 (12%) all the teachers were telling me
they were pull- putting out people’s eyes
3.5 Summary
Because of the empirical challenges relevant to determining correspondences be-

tween linguistic structure and generality of meaning in English conversational ut-
terances, a decision was made to focus on tokens containing general subjects. The
distribution of these utterances suggests that the predicates co-occurring with
general subjects exhibit properties that also express generality of meaning, such as
higher frequency of present-tense stative and imperfective meanings, and lower
frequency of past-tense predicates.
The rest of this paper presents subjective and intersubjective functions of gen-
eralizations in the conversational data. Section 4 discusses generalizations that
function as evaluations in the conversations, and Section 5 illustrates utterance
types that are used to strengthen speakers’ claims.
4. Generalizations as shared evaluations
4.1 Subjectivity, evaluation, and generalization
Linguistic subjectivity is the general phenomenon that refers to the expression of

self and the speaker’s point of view in discourse (e.g., mental states, affect, evalu-
ation, preference). Subjective expressions are those which, to varying degrees, are
grammatically, lexically, or pragmatically anchored to the speaker. Benveniste ob-
serves that “[l]anguage is marked so deeply by the expression of subjectivity that
one might ask if it could still function and be called language if it were constructed
otherwise” (1971: 225). Verhagen (1995: 116) also suggests that subjective expres-
sion is the most usual mode of language use, and that even lexical items like ex-
pensive and tall are not solely informative. Relevant to the current study, research
based on conversational data has noted a variety of associations between linguis-
tic structure and expression of point of view (e.g., Kärkkäinen 2003; Scheibman
2000, 2002; Tao 2001; Thompson and Mulac 1991).
A common type of linguistic subjectivity is evaluation, or the expression of
attitude. Thompson and Hunston (2000) observe that evaluation is conveyed both
lexically (e.g., adjectives such as ugly) and grammatically (e.g., the use of the pro-
gressive in some contexts). Not only are evaluations expressive components of
speakers’ activities, they also play a role in participant interaction. For example, in
conveying opinions, evaluations invariably reproduce cultural norms and values:
“Every act of evaluation expresses a communal value-system, and every act of
evaluation goes toward building up that value-system” (Thompson and Hunston
2000: 6). The generalizations illustrated in this section are evaluations that fulfill
both subjective and intersubjective functions, as characterized by Thompson and
Hunston.
4.2 Construction of an evaluative class
Example (4) is part of a longer conversation among four women friends. Preceding
this episode, one of the participants, Z, a 32-year-old woman, was complaining to
the group that she keeps finding gray hairs, and because they bother her, she pulls
them out whenever she sees them. One of the participants, N, mocks Z about her
worry, characterizing her as wanting to remain a teenager (line 1). The utterance in
line 9, the whole thing is pretty ghastly, is the evaluative generalization in focus. It is
offered by participant O in response to Z’s narrative of her gray hair dilemma.
(4) 1 N: she wants to be a teenager.

2 Z: no I don’t want to be a teenager,
3 but,
4 ...(1.5) no it’s,
5 S: well?
6 O: the whole thing,
7 S: you do it the way you do.
8 Z: (CLEARS THROAT)
9 → O: the whole thing is pretty ghastly.
10 …(1.4) let’s face it that,
11 you know there isn’t much alternative.
12 Z: no,
13 I guess not.
14 N: die young.
15 O: <@ right,
16 [exactly @>]
17 Z: [yeah really].
18 yeah.
19 N: boy,
20 that sounds good.
As noted in Section 3.3, general classes referred to with lexical noun phrases in
the conversational data vary in their conventionality. For example, highly insti-
tutionalized classes are referential groups such as nationalities (e.g., brazilians),
professions (e.g., substitute teachers), or common objects and phenomena
(e.g., ceviche). Less conventional are generalizations based on groups that are
indexed locally in discourse whose interpretation depends, not only on lexical
specification, but also on participants’ sociocultural expectations and knowl-
edge of discourse proceedings (e.g., parents of elementary school students in the
speaker’s class, guys in their twenties in the Bay Area).
An example of a highly indexical construction is illustrated by the generaliza-
tion in (4), the whole thing is pretty ghastly. The lexical noun phrase, the whole
thing, is propositionally empty (the head noun is thing) and hyperbolic (the whole
thing), giving the utterance strong expressive force, especially when combined
with the evaluative complement, pretty ghastly. The referent of the noun-phrase
subject, the whole thing, might be glossed as ‘the experience of aging,’ and in-
cluded in the meaning are shared attitudes toward getting older. Even though the
phrase itself has little explicit content, understanding of this indexical subject by
the interactants is apparent, as demonstrated by subsequent ratification and con-
tinuation of the topic in lines 12–20. It is also the case that O’s generalization has
interpersonal function in the conversation. The utterance comes on the heels of
Z’s defense of herself from N’s teenager accusation (lines 2–4), allowing O to show
solidarity with Z at this point in the conversation (a positive politeness strategy).
In summary, the whole thing is a general construct based on Z’s gray-hair
story and others like it in the culture. It linguistically indexes a popular attitude
toward aging shared by the members of this group (that it is undesirable but inev-
itable). Not only does the generalization express O’s own stance, it also functions
inclusively, or intersubjectively, in the interaction in two ways. O’s statement sup-
ports Z’s gray-hair narrative in the interaction by meeting it with her own nega-
tive evaluation of aging. And at a more global level, O’s generalization expresses a
shared cultural norm. This attitude toward aging is common ground among the
participants (Clark 1996), and one that unifies their individual stances.
4.3 Local and global aspects of intersubjectivity in discourse
Analysis of the preceding example suggests that generalizations have social uses
in friendly English conversations: they mark solidarity between participants, and
they can be sites for the sharing of cultural beliefs. These two functions are aspects
of intersubjectivity. For Benveniste (1971) not only is language inherently subjec-
tive, it is also intersubjective. He writes that “every utterance assum[es] a speaker
and a hearer, and in the speaker, the intention of influencing the other in some
way” (Benveniste 1971: 209). In fact, Benveniste suggests that the process of com-
munication itself is “only a mere pragmatic consequence” of the dialogic relation-
ship between ‘I’ and ‘you’ (1971: 225). Traugott and Dasher (2002: 22) suggest that
“intersubjectivity is most usefully thought of in parallel with subjectivity: as the
explicit, coded expression of SP/W’s [speaker/writer’s] attention to the image of
‘self ’ of AD/R [addressee/reader] in a social or an epistemic sense, for example,
in honorification”. Traugott and Dasher also characterize as intersubjective those
utterances for which the speaker pays particular attention to the addressee as a
speech-act participant (e.g., hedges, politeness markers). Indeed, despite the ana-
lytical convenience of labeling linguistic activities as either subjective or intersub-
jective, as demonstrated in subsequent analyses in this paper as well as in other
chapters in this volume (especially those by Du Bois, Kärkkäinen, Keisanen, and
Haddington), in situations of language use expressions of speakers’ evaluations
and perceptions are fundamentally sensitive to the interactional exigencies of the
context.
Schiffrin (1990) suggests that intersubjectivity is interactively constructed
among participants (e.g., in negotiating turns and topics), and she notes that it
plays a role in the sharing of background knowledge in conversational interac-
tion – activities that make reference to norms and expectations. In a study of evi-
dential expression based on spoken and written Dutch and German, Nuyts (2001)
uses intersubjectivity in a similar way – to refer to people’s sharing of beliefs and
conclusions with the speaker. In these treatments, intersubjectivity refers to shar-
ing of sociocultural expectations.
One characterization of intersubjectivity in discourse, then, focuses on the lo-
cal activities of participants in interactive contexts, and another refers to the shar-
ing of beliefs and attitudes among participants as members of communities. The
former might be characterized as interpersonally, or interactively, intersubjective;
the latter makes reference to more global sharing of expectations. In a study of the
cultural standing of opinions in discourse, Strauss (2004) calls these interpersonal
elements of the context, social (e.g., politeness) and the beliefs and values com-
ponents, cultural. In conversation, however, both of these intersubjectivities are
accomplished simultaneously. Generalizations can be interpersonally inclusive
in that they demonstrate participants’ solidarity or empathy with one another’s

stance. At the same time, because generalizations often make reference to societal
norms, they are involved in the reproduction of belief systems.
4.4 Joint construction of stance using generalizations
In the conversation in (5), three participants are discussing a coworker of speaker

B, who is a woman named Vanda. Specifically the group is trying to figure out
why Vanda is conservative and opinionated, which is how she was characterized
previously in the discourse. The participants agree that Vanda’s having lived in
Vienna until she was 16 could have contributed to her conservative outlook (Vi-
enna is awfully conservative). Even so, the interactants suggest that she should
have more sophisticated views about feminism than she does given that she has
been exposed to a range of cultural experiences due to having lived in different
countries (line 1). There are two generalizations in this conversational extract that
contribute to interactants’ construction of shared stance in discussing this topic:
the jointly produced generalization in lines 2–3 (I would think that a person would
be more/you mean more cosmopolitan), and the generalization offered by speaker
A in lines 6 and 8 (some people are more open to other things than others).
(5) 1 B: .. she’d been exposed to enough that I would’ve --

2 → I would think [that a person would be more],
3 C: [you mean more cosmopolitan].
4 B: yeah.
5 A: but that’s probably her personality,
6 → some people are more open to,
7 C: X
8 A: other things than= .. than others.
The evaluative generalization in lines 2–3 is collaboratively produced. The joint

production of this generalization is accomplished in two ways: the turns of the
two participants are overlapping, and speaker C completes speaker B’s comple-
ment (Ono and Thompson 1995). The collaboration of interactants in articulating
this generalization, along with speaker B’s ratification of C’s completion of her
utterance (line 4), demonstrate the participants’ local construction of a shared at-
titude in the conversation. On the other hand, Speaker A’s generalization in lines
6 and 8 is a contribution to the group’s explanation for Vanda’s attitudes.
Both of the generalizations in this episode (lines 2–3 and lines 6 and 8) contain
generic subjects (a person, some people) co-occurring with relational predicates.
These subject types permit global assessments of people and their behaviors. The
predicate-adjective constructions found in both generalizations are evaluative, in

the sense that notions of what it means to be more cosmopolitan or more open are
subjectively and intersubjectively construed. The shared belief informing the gen-
eralization in lines 2–3 is the expectation that when one lives in different coun-
tries one acquires a certain sophistication. The commonsense view underlying
the generalization in lines 6 and 8 (including line 5) is that people’s attributes can
be divided into those that have to do with cultural experiences and those given
by personality. These generalizations are offered by participants as they collab-
oratively construct an explanation for Vanda’s character – a person unknown to
two of the participants in the conversation. In this way, what began as speaker B’s
subjective stance becomes shared and elaborated by the group.
4.5 Evaluating others with they
The evaluations discussed here are utterances with non-personal, or generic, they
subjects. Because third-person pronouns make reference to non-speech-act par-
ticipants (Quirk et al. 1985: 3354), they is often used by speakers of English and
other languages to exclude, or other, social groups both in public discourse and in
conversational interaction (e.g., Hongladarom 2002 for Thai; Skarżyńska 2002 for
Polish). In English conversations, they is commonly used to evaluate groups that
participants do not belong to, and often these assessments express disapproval or
derision. The use of non-personal they is so conventional in English conversations
(e.g., to refer to institutional entities) that it can occur without its having been
referentially introduced (Biber et al. 1999: 331). For example in (6), the referent of
they was not previously established in the discourse. In this humorous conversa-
tion, they refers to an unnamed but culturally shared conception of an unspeci-
fied institutional entity who is being ridiculed by the participants for engaging in
genetic engineering of chickens for profit.
(6) (Conceptual Pesticides SBC0003 295.15-305.58)

1 ROY: they’re trying to breed like a forty foot long tube chicken?
2 MARILYN: No= @@.
3 You mean so that they can go right from that to Chicken
McNuggets?
The evaluative generalization in the next example (7, line 2) is an utterance con-
taining a generic they subject occurring with a habitual predicate (they come in with
that attitude). The general class designated by non-personal they in this utterance is
substitute teachers. The antecedent for the pronoun occurred a minute before
this point in the conversation when one of the participants, Sharon, tells how a sub
at the public school where she works lost a student’s free-lunch application, which
prevented the child from receiving her lunches. Subsequently, the participants be-
gin to condemn substitute teachers as a group for their general incompetence (they
don’t care; they don’t give a shit about anything). One of the participants, Carolyn,
initiates a second diatribe about beginning substitute teachers, who, according to
her, are more enthusiastic than the experienced group (too enthusiastic, in fact),
but incompetent nonetheless. In the excerpt in (7), Carolyn acts out her evaluation
of a too-eager substitute teacher walking into a classroom.
(7) (Raging Bureaucracy SBC0004 379.05-385.13)

1 CAROLYN: You know,
2 → they come in with that .. attitude,
3 and they go,
4 ((THUMP)) <Q I’ve always wanted to teach math.
5 SHARON: @@
6 CAROLYN: Now,
7 what are we on Q>?
substitute teachers is a relatively conventionalized class in mainstream Amer-

ican English-speaking cultures. Even so, as it is referred to in this conversation,
the class is not simply a referential set of teachers who cover classes on a tempo-
rary basis when “real” teachers are ill or on leave. Rather, important aspects of the
meaning of this class depend on shared stance. That is, they here indexes a class
of people with social and institutional relevance who are viewed derisively by the
conversational participants. The simple form of come in the generalization in line
2 expresses habitual meaning that generalizes over supposed specific instances,
thereby intensifying the speaker’s expression by portraying the behavior she’s sati-
rizing as repetitive. This othering of substitute teachers continues after this point
in the conversation with other participants laughing and joining in the perfor-
mance. In this example, evaluation and elaboration of meaning is accomplished
by all of the participants who share the cultural stance that views substitute teach-
ers as other, creating solidarity among themselves.
4.6 Summary
The examples in this section illustrate situations in which participants’ evaluative

stances expressed with generalizations become jointly held in the conversations.
It is notable that the subjective and intersubjective uses of these utterances are not
distinct from one another. For example, while speaker O’s construction and eval-
uation of a highly indexical class in (4) (the whole thing is pretty ghastly) reflects
her individual stance, the use of this generalization simultaneously demonstrates

intersubjective functions. It is offered to show solidarity with speaker Z’s evalua-
tion of aging, and at a more global level of intersubjective activity, the indexicality
of the subject noun phrase co-occurring with an evaluative predicate expresses a
common cultural view that invites, or perhaps marks, shared interpretation by the
participants. In the excerpt in (5), on the other hand, the relationship between in-
dividual and interactive stance is illustrated by the linguistic co-construction of a
general evaluation based on shared beliefs, motivated by positive politeness (e.g.,
high involvement, Tannen 1984). And finally, generalizations with non-personal
they subjects, illustrated in (6) and (7), are used to mock institutional groups who
are viewed as outsiders. In both of these extracts, speakers offer satirical gener-
alizations that evoke laughter from the other participants, a sign of recognition
as well as amusement (Coates 1996), which suggests that interactants share the
speaker’s evaluation.
5. Augmenting stance with generalizations
5.1 Propositional and interactive broadening
As discussed in the introduction, generalizations exhibit broadening properties.

This broadening can apply to propositional material or to participant interaction,
and both of these aspects can be used to enhance speakers’ stances in conversa-
tions. Propositionally speaking, broadening the domain of applicability of a sci-
entific principle, for example, typically serves to make that principle theoretically
more powerful. Similarly, expanding the reach of an assertion in conversation
using a generalization can potentially increase the expressive power or author-
ity of that assertion. If a speaker’s claim can be considered true for a larger class
of entities or situations, then there is an assumption that the claim is stronger
than it would be, for example, if the assertion were based only on one or perhaps
two specific instances. In this way, generalizations can strengthen a participant’s
stance by appearing to expand the reference class on which a particular claim is
based. Generalizations can also be interactively broadening in the sense that par-
ticipants tend to share beliefs expressed with these utterances (e.g., the whole thing
is pretty ghastly). As illustrated in (9) below, a speaker can explicitly invite other
participants to agree with her generalization by using a generic you pronoun. In
this way, speakers enhance their positions in conversations by appealing directly
to common experiences and beliefs of other participants.
The two generalization types discussed in this section augment speakers’
stances in these two aforementioned ways. The analysis of the generalization in
(8) suggests that this utterance is used by the speaker to reinforce the substance
of his argument by evoking a general group as evidence for his claim. In contrast,
the excerpt in (9) illustrates generalizations with generic you subjects, which are
used by the speaker to strengthen her stance by appealing to the experiences and
beliefs of the other participants. The former is a case in which the broadening
properties of the generalization are applied to the propositional content of the
speaker’s argument, whereas the latter generalizations are interactively expansive
in the sense that they invite corroboration from conversational participants.
5.2 Use of a general class to strengthen a claim
Example (8) is part of a longer conversation in which the speaker, Miles, is incred-
ulous that HIV/AIDS cases in the San Francisco Bay Area had been increasing
dramatically despite the availability of information on preventing transmission of
the disease. A component of Miles’s explanation for this alarming state of affairs
is that women who are having sex with guys in their twenties are not insisting that
these men use condoms. The full generalization is in lines 4–5 (I mean these guys
say these women don’t care if they use rubbers or not), and the antecedent for these
guys appears in lines 1–2 (guys in their twenties).
(8) (Lambada SBC0002 696.49-70594)

1 and just talking with guy=s,
2 ... in their twenties,
3 why,
4 → I mean these guy=s,
5 say these women don’t care if they use rubbers or not.
6 this is what guy=s have been telling me,
Miles’s discursive construction of the demographic, men in their twenties, po-

tentially strengthens his stance in this conversation by broadening the reference
class on which his claim is based. At the same time, he also mitigates having direct
knowledge of the women who are not insisting on using condoms during sex with
the men by using a reported speech construction in lines 5–6, I mean these guys
say these women don’t care if they use rubbers or not (Fox 2001; Hill and Irvine
1993). In the language of Goffman’s (1981) participation framework, Miles is ani-
mating a message that is authored by these guys about these women, and neither
of these general groups (men in their twenties and women they have sex
with) are individual agents whose experiences might be contested.
The other participants in the conversation do not question Miles’s general-
ization. In fact, they provide supportive uptake, such as right, yeah, stupid, and
participate with the speaker in elaborating his claim.11 Subsequent to the excerpt
reproduced in (8), Miles provides evidence for his generalization. He tells the
group that a woman whom he had planned to ask out on a date was telling him
that her sister just won’t use rubbers. Another participant, Jamie, adds her support
by mentioning that her sisters don’t use condoms either. In this interaction, then,
the general argument proposed by Miles – that young women in the Bay Area are
not requiring their male sex partners to use condoms – is accepted, specified, and
jointly maintained by the participants.
5.3 Appealing to other participants
Unlike lexical noun-phrase and they subjects, which index third-person classes
and groups, generic you pronouns formally make reference to speech-act partici-
pants. On the one hand, this pronoun is used by speakers as an informal version
of the generic pronoun one. On the other, it seems to retain second-person refer-
ence in that “the speaker is appealing to the hearer’s experience of life in general,
or else of some specific situation” (Quirk et al. 1985: 354). Built into the speaker’s
generalization using this form, then, is implicit reference to an addressee. The ex-
ample in this section illustrates a frequent context of generic you generalizations
in the database – one in which speakers shift from using first-person-singular
expressions to you forms to augment their stance by appealing to the experiences
and beliefs of the other participants.
The episode in (9) occurs subsequent to a narrative by the primary speaker,
Joanne, in which she described her brother’s down-and-out life before he kicked a
disruptive drug habit. The generalizations in focus are those in Joanne’s final turn
(lines 14–20), But you have to at one point let go. You can’t constantly be torn, just
torn to pieces by, you know, somebody like that.
In the opening line in this excerpt, I mean the guy is great, Joanne is referring
to her brother as he is now since his successful recovery. After this utterance,
Lenore and Ken contribute turns (lines 2–3, 7–8, 13) overlapping with Joanne’s
utterances (lines 1, 4–6, 12) in which they remind her that they had tried to reas-
sure Joanne during the time of her brother’s difficulties that he would be fine (e.g.,
I told you, see).
(9) (Deadly Diseases SBC0015 835.590-850.745)

1 JOANNE: I mean [the guy is] grea=t.
2 LENORE: [I told you].
3 [[I tol=d you=]]
4 JOANNE: [[Isn’t that--
5 he’s not]] hopeless.
6 [Well I had given] him up [for dead].

7 LENORE: [I told you].
8 KEN: [I] always tol=d her,
9 JOANNE: [Oh=,
10 KEN: [it’s gonna,
11 and she’s]--
12 JOANNE: I’d given him up] [for dea-]--
13 LENORE: [see].
14 JOANNE: But you have to --
15 you have to at one point let go.
16 you can’t constantly be torn,
17 just to=rn to pie=ces,
18 by,
19 you know,
20 somebody like that.
21 LENORE: (TSK) So your mother’s happy now.
Notice that in line 14 Joanne switches from using I to you. Based on spoken and
written narratives, Abney (1996) finds that these shifts can be used as distanc-
ing techniques (e.g., from uncomfortable topics). Joanne’s shift from I to generic
you in this conversation may in part be motivated by Lenore and Ken’s repeated
reminders that they had told Joanne all along that her brother would rebound.
These reminders mark a contrast between Joanne’s friends’ expressions of confi-
dence in her brother’s situation and her own lack of confidence, expressed in lines
6 and 12 (I’d given him up for dead). Joanne’s shift to generic you, then, may be
viewed as an iconic distancing from a situation that might hold conflict for her.
Another use of shifts to generic you is to generalize the speaker’s experience
in order to build empathy with other participants. Joanne’s generic you utterances
may justify her previous distancing from her brother by generalizing her reac-
tions for Lenore and Ken. In generalizing her individual stance, she implicitly ap-
peals to societal norms, and by extension, to the beliefs of other participants. This
illustrates the broadening quality of the generalizations. Specifically, the speaker’s
generalizations, But you have to at one point let go. you can’t constantly be torn, just
torn to pieces by, you know, somebody like that, make tacit reference to discourses
of drug and alcohol abuse that advise against codependency. Notice, too, that
these three you utterances all contain the modal elements can’t and have to, which
contribute to their having normative function.
And finally, these general statements expressed with you generalizations in
the conversational data often prompt speaker changes (illustrated in example (9),
line 21). This suggests that generic you utterances fulfill the ritualized intersubjec-
tive function of codas in conversational narratives – expressions conveying “gen-

eral observations that are timeless in character” (Labov and Fanshel 1977: 109).
5.4 Summary
The examples in this section illustrate two ways in which the broadening charac-
teristics of generalizations can potentially strengthen a speaker’s stance in conver-
sations. In (8), the interactive success of Miles’s having indexed a general class to
support his argument can be gauged by the subsequent uptake and specification
of the generalization by participants. Similar to the examples discussed in Section
4, the speaker’s stance in this episode is jointly maintained. In (9), on the other
hand, the speaker generalizes motivations for her past reactions to her brother
using generic you constructions, perhaps to invite her friends to understand, or
agree, with her stance at that point in time. Additionally, as was suggested for the
evaluative generalizations in Section 4, Joanne’s you generalizations are normative
statements that make reference to societal discourses.
6. Summaries and conclusions
The set of conversational generalizations examined in this study are utterances

that designate classes of experiences or entities for purposes of predication, typi-
cally evaluation. Analyses offered in Sections 4 and 5 suggest that these expres-
sions are not referential descriptions of the world. Rather, these utterances dem-
onstrate both subjective and intersubjective uses, and are used by participants in
the conversations to express individual, interactive, and sociocultural stances.
The generalizations are subjective in several ways. Selecting and indexing a
general class as a starting point for evaluation or assertion is invariably an ex-
pressive act, particularly when the class is not a conventionalized one (e.g., the
whole thing). The generalizations in these data frequently function as evaluations
and can also be used as rhetorical devices to strengthen speakers’ stances. Ad-
ditionally, because generalizations can have a broadening, or inclusive, function
in these conversations, they participate in intersubjective activities at an interper-
sonal level (e.g., politeness, demonstrations of solidarity) and more globally in the
maintenance of cultural norms through tacit sharing of societal discourses.
Notably, the subjective uses of generalizations are tied to their intersubjective
uses in several ways. For example, the generalizations with they subjects illustrated
in Section 4 convey speakers’ evaluative stances as they function to create ingroup
solidarity by othering outsiders. In (9), the speaker, Joanne, uses you generaliza-
tions to broaden her stance in an appeal to the beliefs of her friends. In the guys
in their twenties example in (8) and in the extract about Vanda in (5), on the other
hand, participants collaborate with speakers by specifying and elaborating their
generalizations. These analyses support the idea that in English conversations,
individuals’ expressions of stance using generalizations have interactive conse-
quences. And in friendly conversations among intimates a common consequence
is the construction of jointly held stances.
Generalizations in these conversations, then, are expressions that link indi-
vidual subjectivities to the group. Moreover, because these utterances are often
statements of norms, they contribute to the construction and reproduction of
cultural belief systems. Sally McConnell-Ginet observes that “[t]he reproduc-
tion of meaning refers to our dependence, in producing meanings, on previous
meanings or interpretations, to our dependence in particular on one another’s
experience with the linguistic forms being used” (1998: 199). And conversations
are important sites in which participants engage with “one another’s experience,”
both socially and linguistically. Indeed, the generalizations discussed in this paper
allude to common societal discourses such as: the sexual responsibilities of young
women in heterosexual relationships, appropriate ways to engage with people
with drug dependencies, and attitudes toward aging, work, and scientific authori-
ties. Participants’ uses of these generalizations, then, illustrate ways in which links
between expression of individual and social attitudes are reinforced in conversa-
tional interactions.
Notes
1. I am grateful to Robert Englebretson for many valuable comments on an earlier version of

this paper.
2. Langacker describes the reference point phenomenon as the cognitive ability “to invoke the
conception of one entity [reference point] for purposes of establishing mental contact with an-
other [target]” (1993: 5). The strong relational aspects of stancetaking suggest that the reference
point notion might also be a useful way to characterize participant stances in discourse.
3. Language-acquisition researchers have also found that generalizations are sensitive to local
patterns of use. For example, Tomasello (2000) presents compelling evidence from studies us-
ing nonce words suggesting that young children do not generalize verbs as a syntactic category
as they do nouns.
4. In definitions of generalization used in formal systems, such as software and database de-
sign, reduction in specificity of an entity or description is called abstraction. “Abstraction aims
at simplifying the description of an entity, while generalization looks for common properties
among these abstractions” (Kafura 1998: 27). Because my primary objective is to examine gen-
eralizations in English conversations (for which differences between abstraction and generaliza-
tion are not specifically relevant), I will not pursue this technical distinction here.
5. Transcripts of conversational excerpts used in this paper have been edited for readability.
6. A key to transcription symbols is provided in the Appendix.
7. Cats in this conversation refers to a brand of tractor made by Caterpillar.
8. Two general subject types, which were previously addressed, were excluded from the study:
(1) we subjects that designate classes (Scheibman 2004), and (2) utterances with impersonal it
and that subjects (Scheibman 2002 for it and that subjects of relational clauses).
9. Excerpts from the Santa Barbara Corpus of Spoken American English are cited, while those
from unpublished transcripts do not have an accompanying reference.
10. In the conversational extracts, generalizations and utterances that figure prominently in
discussion in the text are bolded.
11. These contributions are not reproduced in (8).
References
Abney, L. 1996. “Pronoun shift in oral folklore, personal experience and literary narratives, or
‘What’s up with you?’” SECOL Review 20: 203–226.
Benveniste, E. 1971. Problems in General Linguistics. Translated by Mary Elizabeth Meek. Coral
Gables, FL: University of Miami Press.
Berman, R.A. 2005. “Introduction: Developing discourse stance in different text types and lan-
guages.” Journal of Pragmatics 37: 105–124.
of evidentiality and affect.” Text 9: 93–124.
Biber, D., Johansson, S., Leech, G., Conrad, S. and Finegan, E. 1999. Longman Grammar of
Bouacha, M.A. 1995. “Linguistique théorique et recherche en didactique.” Le Français dans le
Monde 34: 47–52.
Bybee, J. and Pagliuca, W. 1985. “Cross-linguistic comparison and the development of gram-
matical meaning.” In Historical Semantics, Historical Word Formation, J. Fisiak (ed.), 59–
83. Berlin: Mouton de Gruyter.
Bybee, J., Perkins, R. and Pagliuca, W. 1994. The Evolution of Grammar: Tense, Aspect, and Mo-
dality in the Languages Of The World. Chicago: University of Chicago Press.
Chafe, W. 1994. Discourse, Consciousness, and Time. Chicago: University of Chicago Press.
Clark, H.H. 1996. Using Language. Cambridge: Cambridge University Press.
Coates, J. 1996. Women Talk: Conversation between Women Friends. Oxford: Blackwell.
American English, CD, Part 1. Philadelphia: Linguistic Data Consortium, University of
Pennsylvania.
Pennsylvania.
wards and M.D. Lampert (eds.), 45–89. Hillsdale, NJ: Lawrence Erlbaum.
Fox, B.A. 2001. “Evidentiality: Authority, responsibility, and entitlement in English conversa-
tion.” Journal of Linguistic Anthropology 11: 167–192.
Goffman, E. 1981. “Footing.” In Forms of Talk, 124–159. Philadelphia: University of Pennsyl-

vania Press.
Goldberg, A.E., Casenhiser, D.H. and Sethuraman, N. 2004. “Learning argument structure gen-
eralizations.” Cognitive Linguistics 15: 289–316.
Hill, J. and Irvine, J. (eds.). 1993. Responsibility and Evidence in Oral Discourse. Cambridge:
Hongladarom, D. 2002. “Discourse about them: Construction of ethnic identities in Thai print.”
In Us and Others: Social Identities Across Languages, Discourses and Cultures, A. Duszak
(ed.), 321–339. Amsterdam: John Benjamins.
Hopper, P.J. 1997. “When ‘grammar’ and discourse clash: The problem of source conflicts”. In
Essays on Language Function and Language Type, J. Bybee, J. Haiman and S.A. Thompson
(eds.), 231–247. Amsterdam: John Benjamins.
Kafura, D. 1998. Object-Oriented Software Design and Construction with C++. Upper Saddle
River, NJ: Prentice Hall.
Kärkkäinen, E. 2003. Epistemic Stance in English Conversation: A Description of its Interactional
Functions, With a Focus On ‘I think’. Amsterdam: John Benjamins.
Labov, W. and Fanshel, D. 1977. Therapeutic Discourse: Psychotherapy as Conversation. New
York: Academic Press.
Labov, W. and Waletzky, J. 1967. “Narrative analysis: Oral versions of personal experience.” In
Essays on the Verbal and Visual Arts, J. Helm (ed.), 12–44. Seattle: University of Washing-
ton Press.
Lakoff, G. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind.
Chicago: University of Chicago Press.
Langacker, R.W. 1987. Foundations of Cognitive Grammar, Vol. I: Theoretical Prerequisites. Stan-
ford, CA: Stanford University Press.
Langacker, R.W. 1993. “Reference-point constructions.” Cognitive Linguistics 4: 1–38.
Lyons, J. 1977. Semantics, Vol. 1. Cambridge: Cambridge University Press.
MacWhinney, B. 1977. “Starting points.” Language 53: 152–168.
McConnell-Ginet, S. 1998. “The sexual (re)production of meaning: A discourse-based ap-
proach.” In The Feminist Critique of Language: A Reader, 2nd ed., D. Cameron (ed.), 198–
210. London: Routledge.
Nuyts, J. 2001. “Subjectivity as an evidential dimension in epistemic modal expressions.” Jour-
nal of Pragmatics 33: 383–400.
Ono, T. and Thompson, S.A. 1995. “What can conversation tell us about syntax?” In Alterna-
tive Linguistics: Descriptive and Theoretical Modes, P.W. Davis (ed.), 213–271. Amsterdam:
John Benjamins.
Quirk, R., Greenbaum, S., Leech, G. and Svartvik, J. 1985. A Comprehensive Grammar of the
English Language. London: Longman.
Reilly, J., Zamoraa, A. and McGivern, R.F. 2005. “Acquiring perspective in English: The devel-
opment of stance.” Journal of Pragmatics 37: 185–208.
Scheibman, J. 2000. “I dunno ... A usage-based account of the phonological reduction of don’t in
American English conversation.” Journal of Pragmatics 32: 105–124.
Scheibman, J. 2004. “Inclusive and exclusive patterning of the English first person plural: Evi-
dence from conversation.” In Language, Culture and Mind, M. Achard and S. Kemmer
(eds.), 377–396. Stanford, CA: CSLI Publications.
Schiffrin, D. 1990. “The principle of intersubjectivity in communication and conversation.” Se-

miotica 80: 121–151.
Silverstein, M. 1976. “Shifters, linguistic categories, and cultural description.” In Meaning in
Anthropology, K.H. Basso and H.A. Selby (eds.), 11–55. Albuquerque: University of New
Mexico Press.
Skarżyńska, K. 2002. “We and they in Polish political discourse.” In Us and Others: Social Iden-
tities across Languages, Discourses and Cultures, A. Duszak (ed.), 249–264. Amsterdam:
John Benjamins.
Strauss, C. 2004. “Cultural standing in expression of opinion.” Language in Society 33: 161–
194.
Tannen, D. 1984. Conversational Style: Analyzing Talk among Friends. Norwood, NJ: Ablex.
Tao, H. 2001. “Discovering the usual with corpora: The case of remember.” In Corpus Linguistics
in North America: Selections from the 1999 Symposium, R. Simpson and J. Swales (eds.),
116–114. Ann Arbor: University of Michigan Press.
Thompson, S.A. and Mulac, A. 1991. “A quantitative perspective on the grammaticization of
epistemic parentheticals in English.” In Approaches to Grammaticalization, Volume II: Fo-
cus on Types of Grammatical Markers, E. Traugott and B. Heine (eds.), 313–329. Amster-
dam: John Benjamins.
Tomasello, M. 2000. “Do young children have adult syntactic competence?” Cognition 74:
209–253.
Traugott, E. and Dasher, R.B. 2002. Regularity in Semantic Change. Cambridge: Cambridge
University Press.
van Leeuwan, T. 1995. “Representing social action.” Discourse and Society 6: 81–106.
Verhagen, A. 1995. “Subjectification, syntax, and communication.” In Subjectivity and Sub-
jectivization: Linguistic Perspectives, D. Stein and S. Wright (eds.), 103–128. Cambridge:
Zhang, Qiao. 1998. “Fuzziness – vagueness – generality – ambiguity.” Journal of Pragmatics 29:
13–31.
Appendix: Transcription symbols
(Du Bois et al. 1993)
Speech overlap [ ] or [[ ]]
Truncated intonation unit --
Transitional Continuity
Final .
Continuing ,
Appeal ?
Lengthening =
Medium pause …
Long pause …(N)
Laughter @
Laugh quality <@ @>
Quotation quality <Q Q>
Vocal noises ()
Indecipherable syllable X
Researcher’s comment (( ))
The stance triangle
John W. Du Bois
University of California, Santa Barbara
1. Introduction1
One of the most important things we do with words is take a stance. Stance has
the power to assign value to objects of interest, to position social actors with re-
spect to those objects, to calibrate alignment between stancetakers, and to invoke
presupposed systems of sociocultural value. Yet very little is understood at pres-
ent about stance: what it is, how we do it, what role language and interaction play
in the process, and what role the act of taking a stance fulfills in the broader play
of social life. These are the questions which will inform this paper, as I seek to
develop a language for describing the phenomenon of stance and clarifying the
role it plays in the larger contexts of language and interaction. If we are to achieve
any general understanding of stance, it will be necessary to seek the foundational
principles which underlie the act of taking a stance and negotiating its meaning.
Because the diversity of observable stances extends in principle without limit, it
is necessary to go beyond merely cataloguing their contents or classifying their
types. To frame a theory of stance means to provide a general account of the mode
of production of any stance and of its interpretation in a context of interaction.
Realizing such a goal will require us to define a research agenda capable of bring-
ing together multiple coordinated lines of inquiry drawn from a range of disci-
plines concerned with the use of language. As one step in this direction, this paper
presents a preliminary sketch of some of the theoretical resources and analytical
tools that are likely to be required for such an account.
Stance can be approached as a linguistically articulated form of social action
whose meaning is to be construed within the broader scope of language, interac-
tion, and sociocultural value. Setting the problem in this way brings into play
several aspects of language in interaction. As we seek the theoretical resources
needed to account for the achievement of stance, we find ourselves faced with a
140 John W. Du Bois
complex web of interconnections linking stance with dialogicality, intersubjectiv-

ity, the social actors who jointly enact stance, and the mediating frameworks of
linguistic structure and sociocultural value they invoke in doing so.
Dialogicality makes its presence felt to the extent that a stancetaker’s words
derive from, and further engage with, the words of those who have spoken be-
fore – whether immediately within the current exchange of stance utterances, or
more remotely along the horizons of language and prior text as projected by the
community of discourse. While the idea of dialogicality has proved theoretically
evocative (Bakhtin [1934]1981; Voloshinov [1929]1973), it cannot be said to have
translated successfully into a well-defined program for substantive research on
observed instances of dialogic interaction (but for some hopeful possibilities see
Linell 1998). In this light, stance represents a promising testing ground to ex-
plore the potential of a more explicit dialogic method in the context of conversa-
tional interaction. Here I draw on my recent formulation of dialogic syntax as a
specifically dialogic method (Du Bois 2001). Dialogic syntax looks at what hap-
pens when speakers build their utterances by selectively reproducing elements of
a prior speaker’s utterance. As an analytical practice, dialogic syntax details the
process of mapping resonances between juxtaposed utterances in discourse. As it
happens, when the formal method of dialogic syntax is applied to conversational
interaction, it turns up a remarkable number of stance pairs characterized by a
recurring functional-interactional configuration, in which the stance utterance of
a subsequent speaker is constituted as bearing a close analogy to the stance utter-
ance of a prior speaker. As stances build on each other dialogically, the analogy
implied by their structural parallelism triggers a series of interpretative and inter-
actional consequences, which will be seen to carry significant implications for the
interaction at hand, and, at a more general level, for the theory of stance.
Injecting dialogicality into the analysis of stance leads naturally to a concern
with intersubjectivity, which I understand as the relation between one actor’s sub-
jectivity and another’s. Though its conceptual origins lie in philosophy, intersub-
jectivity has long been recognized as an issue for the social sciences, where it
has influenced especially those approaches which locate language and cognition
within a sociocultural and/or sociocognitive perspective (Bruner 1986; Clancy
1999; Hanks 1990; Lucy 1993; Ochs 1988; Schutz 1962; Silverstein 2001; Toma-
sello 1999; Vygotsky 1986). It should be noted that intersubjectivity presupposes
subjectivity. While subjectivity has been getting the greater share of attention late-
ly, being recognized for its role in the cognitive organization of language structure
(Langacker 1991; Traugott 1989) and the discourse-functional organization of
language use (Maynard 1993; Scheibman 2002), intersubjectivity is no less indis-
pensable as a piece of the larger stance puzzle. The significance of the necessary
link between the intersubjective, subjective, and objective will become evident
once we recognize them as fundamental sociocognitive relations which organize

language use. Intersubjectivity and its companions play a prominent role in the
present approach to stance, insofar as they serve to ground the sociocognitive
aspects of stancetaking in dialogic interaction.
To complete the picture, we need to look at how the dialogic and intersubjec-
tive dimensions of stance relate to the stancetakers’ actions, and to the sociocul-
tural frames that mediate the consequences of their actions. Stance is realized, in
the usual case, by a linguistic act which is at the same time a social act. The act of
taking a stance necessarily invokes an evaluation at one level or another, whether
by assertion or inference. This in turn implicates those dimensions of sociocul-
tural value which are referenced by the evaluative act. Sociocultural value is mo-
bilized and deployed through stance processes. Via specific acts of stancetaking,
value can be focused and directed at a precise target, as locally relevant values are
activated to frame the significance of participant actions. Moreover, since lan-
guage is reflexive (Haviland 1996; Lucy 1993), and participants routinely monitor
who is responsible (Hill and Irvine 1993) for any given stance, the very act of tak-
ing a stance becomes fair game to serve as a target for the next speaker’s stance.
Stance both derives from and has consequences for social actors, whose lives are
impacted by the stances they and others take. Moreover, in many cases the cur-
rent stance act resonates both formally and functionally with a stance taken in
prior discourse. Thus the value of any stance utterance tends to be shaped by its
framing through the collaborative acts of co-participants in dialogic interaction.
We begin to appreciate why stance should come to wield both subtlety and power
in the dynamics of social life.
In this paper, I work to bring these elements together in a unified framework
for stance. The model I propose is articulated in terms of a set of triangular rela-
tions among the components of stance. But before we get to the stance triangle,
there is groundwork to be laid. I begin the paper with a brief presentation of sev-
eral different kinds of stance phenomena. I then address the problem of contextu-
alizing stance, articulating a set of questions designed to identify some of the key
aspects of the indexical context of interaction which are needed for interpreting
stance. I go on to explore the role of intersubjectivity and other sociocognitive
relations in the elaboration of stance, arguing that making sense of intersubjectiv-
ity in interaction depends on a dialogic understanding of language use. This sets
the stage for the next level of theorizing stance. I introduce the stance triangle as
a way of representing the components of the stance act, and more importantly,
articulating their multiplex interrelations. In its guise as a representation of the
fundamental structure of the stance act, the stance triangle attempts to shed light
on the realization, interpretation, and consequences of stance in interaction. The
result is a fundamentally dialogic perspective on stance which sees the stance act
142 John W. Du Bois
as shaped by the complex interplay of collaborative acts by dialogic co-partici-

pants. By way of conclusion, I summarize the main contributions of the stance
triangle, and pose the question of what, after all, is so special about stance.
2. Kinds of stance
One way to begin thinking about stance is to look at some likely exemplars, based
on what has been recognized as stance (or a stance-related category) by one ana-
lytical tradition or another. While the range of proposals in the literature is too
broad to survey here, what we can do is look at a few of the types that have played
a leading role in previous discussions of kinds of stance or stance-related catego-
ries. For now we can think of these as different kinds or types of stance acts, but
in the end we will have to consider alternative formulations of the issue of stance
diversity. By the same token, although for expository purposes we begin with ab-
breviated examples of stance utterances viewed in isolation, it will soon become
clear that the actual stance taken cannot be fully interpreted without reference to
its larger dialogic and sequential context.
Perhaps the most salient and widely recognized form of stancetaking is evalu-
ation. Evaluation has received considerable attention in recent years (Conrad and
Biber 2000; Hunston and Sinclair 2000; Hunston and Thompson 2000; Labov
and Waletzky 1967; Lemke 1998; Linde 1997; Macken-Horarik and Martin 2003;
Thompson and Hunston 2000). A closely related concept is that of assessment, as
analyzed in conversation analysis (Goodwin and Goodwin 1992; Goodwin 2006;
Pomerantz 1984). Work on the related notion of appraisal has been pursued from
the perspective of systemic functional grammar (Martin 2000), and additional
important work on stance, point of view, and related notions has been developed
by a number of scholars (Berman et al. 2002; Berman 2005; Chafe 1994; Kärk-
käinen 2003a, 2003b; Kockelman 2004; Shoaps 2004).
Consider the following three examples of evaluation, drawn from the Santa
Barbara Corpus of Spoken American English (Du Bois et al. 2000; Du Bois et al.
2003):2, 3
(1) (Conceptual Pesticides SBC003: 317.46-318.16)

PETE; That’s horrible.
(2) (Runway Heading SBC022: 193.440-194.341)
LANCE; (TSK) That’s ideal.
(3) (Appease the Monster SBC013: 1117.12-1118.12)
KEVIN; @that’s @nasty.
In these examples, taken from three different conversations, the stance predicates
horrible, ideal, and nasty are used to evaluate something. The thing evaluated –
the specific target toward which the evaluation is oriented – is referred to in each
case by the demonstrative pronoun that. The evaluative target may be called the
object of stance (for reasons that will become clear in due course). In general
terms, evaluation can be defined as the process whereby a stancetaker orients to
an object of stance and characterizes it as having some specific quality or value.
The next set of examples presents a somewhat different pattern of stance:

LANCE; I’m glad.
(5) (Hey Cutie-Pie SBC028: 52.330-53.355)
JEFF; (TSK) I’m so glad.
(6) (Lambada SBC002: 665.79-667.35)
MILES; I’m just ama:zed.
In these stance utterances, the first person pronoun I, indexing the stancetak-
er, is followed by an affective predicate, positioning the speaker as glad, so glad,
or amazed.4 As speakers position themselves affectively, they choose a position
along an affective scale – as, for example, either glad or so glad. Such utterances
have often been described as indexing affective stance (Besnier 1993; Haviland
1991; Maynard 1993; Ochs 1996; Shoaps 2002).
Speakers may position themselves not only along an affective scale, but also
along an epistemic scale (Clift 2006; Haviland 1991; Heritage and Raymond 2005;
Kärkkäinen 2003b; Thompson and Mulac 1991), for example, presenting them-
selves as knowledgeable or ignorant:

KENDRA; I kno:w.
(8) (Risk SBC024: 588.565-588.970)
DAN; .. I don’t know.
The general concept which subsumes both affective and epistemic stance acts of
the sort illustrated here is positioning (Davies and Harré 1990; Du Bois 2002a).
Positioning can be defined, provisionally, as the act of situating a social actor with
respect to responsibility for stance and for invoking sociocultural value. In the
examples presented here for both affective and epistemic positioning, the speaker
who is taking the stance is indexed via a first-person pronoun in syntactic subject
role (I), while the stance predicate (adjective or verb) specifies the nature of the
stancetaker’s position, whether with respect to an affective (glad) or an epistemic
(know) state, or both (amazed).
144 John W. Du Bois
In the next set of examples, the structural pattern is superficially similar to

the one just presented:
(9) (LSAC 1296-02)5

PAT; I agree.
(10) (Doesn’t Work in This Household SBC019: 127.391-128.183)
MELISSA; I totally agree.
Here a first-person pronoun in syntactic subject position is followed by a stance

predicate (the verb agree). But what is being achieved thereby is different enough
to warrant recognition as a distinct type of stance. The difference lies in the prag-
matic-interactional configuration that it enacts. By uttering I agree the speaker
defines her stance in relation to that of another party, typically the person she is
addressing. The addressee’s role is usually left implicit, but can be made explicit
on occasion:
(11) (LSAC 1396-01)6

LESLIE; I agree with you.
As this example illustrates, agreement normally implies agreement with some-

one,7 which in conversation is usually the person being addressed. The general
term for this kind of stancetaking is alignment (Du Bois 2002a; Heritage 2002;
Heritage and Raymond 2005). Alignment can be defined provisionally as the act
of calibrating the relationship between two stances, and by implication between
two stancetakers. Although a stance verb like agree may be the most transparent
way to display alignment with another speaker, it is certainly not the usual way.
More commonly, speakers show alignment by stance markers like yes or no, or
gestures like a nod or a headshake, or any number of other forms that index some
degree of alignment. Just as often, participants allow their alignment to remain
implicit, inviting the listener to infer it based on comparing the relevant stances.
The role of implicit stance alignment will prove to be especially important in the
management of intersubjectivity.
These examples make it clear that stance utterances can vary in significant
and systematic ways. But the question remains as to how to interpret this vari-
ability. Of the cases examined so far, most seem to involve one of three well-differ-
entiated kinds of stance function, namely evaluation (that’s horrible), positioning
(I’m glad), and alignment (I agree). But do these stance functions really represent
separate categories of stance – or simply different facets of the speaker’s stance,
more broadly conceived? What about other candidates that have been proposed
as stance types – epistemic stance (I know), affective stance (I’m glad), and more?
The problem is, while these are sometimes viewed as different types of stance
acts, no one has yet resolved the question of how many such types should be
recognized – whether there will turn out to be three, five, or many more subtypes
of stance. Nor have the criteria for deciding the question been established. More
seriously, there is the matter of whether the various forms of stance occur sepa-
rately or together. In other words: Can a single stance combine aspects of more
than one of these stance functions? This question raises the possibility of an alter-
native approach which recognizes a more complex picture of the stance act, seen
as encompassing multiple facets at once. Building a more unified understanding
of stance has the advantage of avoiding a limitless proliferation of stance types.
Such an approach would seek to interpret the diversity of stances not as distinct
types of stance, but simply as different facets of a single unified stance act. The
choice between proliferation of stance types and integration of stance facets will
be returned to below.
The answer to some of these questions may hinge in part on the realization
that, in the analysis presented so far, something seems to be missing. In stances
which were characterized as positioning (4–8), for example, the question arises:
What is the speaker positioning himself about? Similarly, in stances involving
alignment (9–11), it seems equally essential to ask: What is the speaker agreeing
about, and who are they agreeing with? These questions point to some important
issues about stance. Speakers do not just perform generic stance types, they per-
form specific stance acts, which have specific content and are located in a particu-
lar dialogic and sequential context. To fully understand what stance is being taken
on any given occasion, we need to learn more about how speakers realize stances
and how hearers interpret their situated meanings. In part, this question con-
cerns how participants contextualize the stance utterance in order to interpret it,
as the stance emerges across successive utterances through processes of dialogic
action. This is the issue we begin to explore in the next section.
3. Contextualizing the emerging stance
If stance is an act, we should expect to locate it in utterances. For the sake of ar-
gument, however, let us begin this discussion where it has so often begun before
in the study of language – not with the utterance, but the sentence. The sentence
can, with a little imagination, be hypothesized as an abstract linguistic structure
detached from any mooring in a specific context of use. What can the decontextu-
alized sentence reveal – and what can it not? Consider the following sentences:
(12) The Caribbean is incredible.

(13) It was really great.
146 John W. Du Bois
(14) I would love to go.
Each of these sentences contains at least one apparently evaluative word: incred-
ible, great, and love (and perhaps others). These words are obviously not neutral
descriptions of external reality, but imply value judgments regarding some refer-
ent. The evaluative connotations of such words are evident even from sentences
taken in isolation. In fact, the evaluative meaning often comes through even in an
arbitrary listing of individual words (incredible, great, ideal, horrible, nasty, and
so on).
But stance is more than the context-free connotations of words or sentences.
In the grammarian’s standard presentation of the isolated sentence, stance re-
mains incomplete. The missing ingredients can only be found by contextualizing
the utterance, defined as the situated realization of language in use. Any utterance
carries cues for its own indexical contextualization (Gumperz 1992; Silverstein
1976). Contextualization cues (or indexical signs) work by pointing beyond the
utterance to its presupposed conditions of use. Thus in situating the understand-
ing of any stance, the first question is: What’s missing? In pragmatic terms this
translates to: What are the indexical absences? Tracing out the salient indexical
meanings (Jakobson [1957]1990; Peirce [1885]1933; Silverstein 1976) helps us to
identify those aspects of context which must become known in order to arrive at
a successful interpretation of the stance at hand.
There are at least three things we need to know about a given occasion of
stancetaking, beyond what may be overtly present in the words and structures
of the stance sentence itself: (1) Who is the stancetaker? (2) What is the object of
stance? (3) What stance is the stancetaker responding to? Each question points to
one component of the process of interpreting stance.
3.1 Who is the stancetaker?
In conversation, participants normally care who says what, and monitor it ac-
cordingly. Participant awareness of the attributability of utterances is routinely
represented in most systems for transcribing discourse (Du Bois 1991; Du Bois
et al. 1993; Jefferson 2004), typically by supplying each transcribed utterance or
turn with a label indicating the identity of the speaker. In the following utterances
by three different speakers, the identity of the speaker is indicated as usual at the
beginning of each turn.
(15) (Deadly Diseases SBC015: 210.075-212.730)

JOANNE; the– the Caribbean is incre:dible.

MARILYN; it was really great.
KEN; I would love to go:.
One important difference between attested utterances like those in (15–17) and
hypothetical sentences like those abstracted in (12–14) is that a real utterance is
always framed by its context of use. A key component of the context of any utter-
ance is the speaker who is responsible for it. In the representations of utterances
given here, we are told that someone named Joanne8 is responsible for the claim
that a certain place is incredible; that Marilyn is responsible for the statement
that something was really great; and that Ken is responsible for a display of affect
or preference about going somewhere. To be sure, just attributing speakership in
this way doesn’t reveal much unless the speaker’s identity carries some significant
associations for us. But in most conversational settings, participants may draw on
a range of biographical associations for the current speaker, regardless of whether
they happen to know the speaker’s name. Participants remember interactionally
salient information about co-participants, and so may factor into their stance in-
terpretation, in addition to what the speaker is saying right now, some or all of
the following: what the speaker has said previously (whether on this occasion or
some other); what sort of relationship they have displayed up to now relative to
co-present others; what accent, voice quality, and intonation they are speaking
with; what their displayed regional, ethnic, gender, or other identities may be;
whether they appear entitled to their claimed identities; details of their life story,
if known; and so on. Moreover, participants derive memorable information about
each other from stances taken, which they may retain as socially salient, with the
option of introducing it into future processes of stance interpretation. Knowing
which social actor is responsible for a specific stance utterance in the past can
make a significant difference in the interpretation of a current stance utterance, in
part because of the dialogic connections that arise between stances.
3.2 What is the object of stance?
To make sense of a given stance we need to know not only who is speaking, but
what they are speaking about. Among other things, we need to know the referen-
tial object or target toward which the stance is being directed – for example, what
is claimed to be incredible or great, where the speaker displays a desire to go, and
so on. Consider the following examples:
148 John W. Du Bois
(18) (Raging Bureaucracy SBC004: 17.67-18.82)

CAROLYN; That stuff is great.
MARILYN; it was really great.
If two people each evaluate something as great, have they taken the same stance?
The cited utterances, drawn from two different conversations, differ slightly in
their evaluative predicates (is great vs. was really great) and in the noun phrases
which designate their respective objects of stance (that stuff vs. it). But these small
differences don’t really speak to the main question. What we want to know, if we
are to decide what stance is being taken, is what it and that stuff refer to. A crucial
part of interpreting any stance utterance is identifying the object of stance, as part
of the process of referential grounding (Hanks 1990). Often the immediate prior
discourse provides sufficient contextualization to resolve the reference of a pro-
noun, demonstrative noun phrase, or other referring form, thus establishing the
identity of the relevant object of stance:

MARILYN; S– really relaxing weekend,
you know,
it was really great.
(21) (Raging Bureaucracy SBC004: 15.42-18.82)
CAROLYN; You know what’s good is .. hibiscus cooler.
(0.2)
That stuff is great.
A weekend and a drink of hibiscus cooler may both be great, but they will be
great in very different ways. The respective stances of Marilyn and Carolyn, once
referentially grounded, cannot be regarded as the same. Stance is a property of
utterances, not of sentences, and utterances are inherently embedded in their dia-
logic contexts.
It’s not just pronouns that need to be resolved by contextualizing the stance
utterance, but also the meaning of content words and other elements as well.
Compare the following two stance utterances, which both employ the word in-
credible in the evaluative predicate, but apply it to rather different stance objects:

JOANNE; the– the Caribbean is incre:dible.
(23) (Cuz SBC006: 168.60-170.30)
ALINA; (H) The male athletes were incredible,
The evaluative content of the message conveyed by an evaluative word like incred-
ible may shift according to whether it applies to a vacation destination (valued for
passive attributes like visual beauty, tranquility, and so on) or to athletes (valued
for the active display of strength, agility, talent, and so on).
In I would love to go (17), the utterance may initially appear more self-con-
tained, as if carrying its meaning complete within itself. But this illusion is soon
dispelled by examination of the prior discourse:

1 JOANNE; (H) I don’t know how many Americans have been
to Nicaragua.
((39 LINES OMITTED))
41 KEN; I would love to go:.
Here, the phrase love to go refers not to going in the abstract, nor to going some-
where at random, but specifically to going to Nicaragua. The stance object that
Ken is evaluating is thus to go to Nicaragua, abbreviated as to go. In such cases,
the contextualized stance utterance takes its interpretation in part from the prior
stance of a dialogic co-participant. This contextualization remains essential to the
interpretation of the current stance even if the relevant prior stance occurred quite
a bit earlier, as is the case here. In sum, identifying the object of stance – what the
evaluation is about – is an essential part of the process of stance interpretation, for
participants and analysts alike.
3.3 What stance is the stancetaker responding to?
Knowing the identity of the stancetaker and the object of stance is a good start,
but we remain on uncertain ground until we know what prior stance the current
stance is being formulated in response to – its counterstance, if you will. Why this
stance is being taken, why just now, why in these terms – to answer these questions
we need to monitor the dialogic and sequential shape of the ongoing exchange of
stance and counterstance. Consider the examples of alignment given previously (I
agree, I agree with you). Despite the fact that as sentences these are grammatically
complete, as stances they are pragmatically incomplete. People don’t agree in the
abstract, they agree with someone about something. While a sentence like I agree
with you (11) foregrounds – or profiles, in Langacker’s (1987) terms – the dimen-
sion of alignment almost exclusively, for its interpretation it must still indexically
incorporate a prior stance content, including the relevant object of stance. Nor-
mally the relevant stance content will be locatable in the prior discourse, specify-
ing what specific stance is being agreed with:
150 John W. Du Bois
(25) (LSAC 1396-01)

1 CORA; we are considered white collar because we’re social
workers.
((2 LINES OMITTED))
4 LESLIE; I agree with you.
Here, Leslie’s I agree with you creates a convergent alignment with her addressee
which immediately entails or implicates an endorsement of the addressee’s stance.
The result can be informally paraphrased9 as I agree with you that we are considered
white collar (because we’re social workers). Similarly in the following example:
(26) (LSAC 1296-02)

1 KIM; No more or no less than any other school is the way I
see it.
2 PAT; Yeah,
3 I agree.
4 I agree.
When Pat says I agree in line 3 (and again in line 4), she is doing more than just
“being agreeable” (or even “doing agreement”). Her stance is quite specific, and
can be paraphrased as, I agree (with you that it’s) no more or no less than any other
school....10 Finally, consider Melissa’s utterance of I totally agree (10), whose rel-
evant context is as follows:
(27) (Doesn’t Work in This Household SBC019: 98.652-129.138)

1 JAN; Take it downstairs.
30 MELISSA; I totally agree.
31 I should go downstairs.
In saying I totally agree, Melissa is not agreeing in the abstract, but specifical-
ly agreeing with her mother Jan’s directive that she should take her homework
downstairs. The requirement to specify the particular content for the stance of
agreement is so strong that participants will go back 29 lines in the conversation,
if need be, to find the proposition spelled out overtly. Lest there be any doubt as
to what proposition she is agreeing to in line 30, Melissa makes it explicit in her
next utterance (line 31), yielding a composite stance that is roughly paraphrasable
as I totally agree (that) I should go downstairs.11 To generalize, given a decontex-
tualized sentence which apparently expresses simple agreement (I agree, I totally
agree, I agree with you), it is not possible to tell from the sentence alone what is
being agreed with.12 Only by referencing the relevant prior stance, locatable ana-
phorically in the dialogic context, can the meaning of the present agreeing stance
be understood.
We have considered three questions about the context of stance which are
likely to be relevant, in some formulation or other, to the interpretation of any act
of stancetaking. In asking, Who is the stancetaker, what is the stance about, and
what stance is the stancetaker responding to? we are seeking to fill in some of the
blanks that must be filled if we want to understand what the stance now being tak-
en actually is. In more general terms, these questions about stance can be linked
to notions of stance subject, stance object, and alignment, connections which will
be further developed below. But we are far from claiming that these are the only
questions, or necessarily the main ones, or the best formulations of them. Cer-
tainly there are additional questions that will need to be posed as we continue our
efforts to tease out the recurrent features of the stance act, and some of these will
come up below. But it is important to point out here that even the questions in-
formally posed so far frequently require participants (and analysts) to go beyond
what is explicit in the words of the stance utterance itself. Still the questions must
be answered. Their relevance to stance does not depend solely on the presence
of explicit words, gestures, prosody, or other communicative elements, however
important these may be, but is grounded ultimately in the systematic knowledge
which participants control regarding what can be expected to be present in any
stance. The constant relevance of the general components of stance influences
what we expect to know about any act of stancetaking, and thereby shapes its spe-
cific interpretation. This holds true whether the information is directly expressed
in the stance utterance, or is only to be found distributed across multiple utter-
ances by different speakers within extended sequences of dialogic exchange. The
claim is that in each case, certain well-defined items of information are actively
sought out by participants in response to the projectable structure of stance.
We have been considering how the contextualization of actual stance utter-
ances (e.g., 15–27) contributes to the interpretation of stance. In contrast, little
or nothing of the relevant contextualizing information can be gleaned from in-
specting idealized sentences (e.g., 12–14) in isolation, even if the sentences are
fully grammatical and fully meaningful – at least as meaningful as they can be, if
limited to the sentential level of referring-and-predicating functions (Silverstein
1976, 2001). Yet once we identify the relevant contextual features, the necessary
ingredients start to fall into place for a full pragmatic and interactional interpreta-
tion. In constructing a logic of stance interpretation, what we want to describe is
the participant’s interpretive process, which we track by close observation of their
own interpretive actions in stance-rich environments. This interpretive inquiry is
akin to that which has motivated scholars as diverse as Geertz (1973) and Sacks
(1992), whose overt theoretical positioning may otherwise appear so disparate as
152 John W. Du Bois
to be incommensurable. We need contextual grounding (Gumperz 1992; Hanks

1990; Silverstein 1976) in dialogic and sequential context (Linell 1998; Sacks 1992;
Schegloff 1996; Voloshinov [1929]1973) to arrive at a sufficiently enriched inter-
pretation of the utterance (Recanati 1989; Sperber and Wilson 1995). We cannot
overlook the influence of even so basic interpretive processes as these if we want
to understand how stances come to be understood. Ultimately, the specificity of
stance is grounded in its dialogicality. Only when the dialogic context is taken
into account does stance become complete.
4. Subjectivity and positioning
In light of this understanding of the processes of stance contextualization, we are

now ready to re-examine the issue of positioning, which we consider in light of
speaker subjectivity. Subjectivity and positioning go hand in hand, inasmuch as
the act of positioning regularly invokes a dimension of speaker subjectivity. Con-
sider the following:
(28) (Lambada SBC002: 865.91)

JAMIE; I like this song.
(29) (This Retirement Bit SBC011: 444.12-445.36)
SAM; I don’t like those.
In these stance utterances, the subjective dimension is registered overtly through

several discrete linguistic elements, including personal pronouns (I) and affec-
tive verbs (like, don’t like). The personal pronoun I points directly to the speaking
subject (Benveniste 1971; Culioli 1990; Ducrot 1972, 1984)13 who, in these cases
at least, is the one taking the stance. The affective stance predicate indexes spe-
cific aspects of the subject’s feelings, positioning the speaker subjectively along
some scale of affective value. Yet it cannot be overlooked that subjective predi-
cates which are transitive (like, don’t like, love, hate, and so on) regularly specify
also the object to which the subject is orienting affectively.14 The transitive clause I
like this song (28), uttered in the auditory presence of a song playing on the stereo,
directly specifies the object of stance – the entity being oriented to. This example
compactly illustrates the co-existence of subjective (I) and objective (this song)
elements within a unified stance. Similarly, in I don’t like those (29), the coherent
expression of speaker subjectivity requires both a subject and an object of stance
(specified as I and those, respectively). In sum, to articulate this kind of subjectiv-
ity, what is required is an orientation to a specific object of the speaking subject’s
stance, combined with specification of a particular intentional relation (Searle

1983), such as desiring, loving, hating, liking, not liking, and so on.
The following stance diagrams represent the relations of evaluation and posi-
tioning between the stance subject and stance object:
(30) Speaker Stance Subject Positions/Evaluates Stance Object

JAMIE; I like this song
(31) Speaker Stance Subject Positions/Evaluates Stance Object
SAM; I don’t like those
This kind of stance diagram is useful in making explicit several aspects of our
analysis. First, we find it essential to represent the identity of the speaker, because
this tells us who is the speaking subject (the stancetaker). Second, in analyzing the
various discrete components of the utterance, we label the words which overtly
express or index the stance subject and the stance object. Third, we label the verb
or other stance predicate according to the kind of stance action it performs. In
these examples, the predicate (like, don’t like) obviously serves to position the sub-
ject, but it also commits the stancetaker to a certain evaluation of the object. In
recognition of this apparent dual stance function, the representations of evalua-
tion and positioning are combined in a single column.
While some stance utterances evidently perform combinations of functions
(for example, evaluation of a stance object combined with positioning of a stance
subject), we might still ask whether it is possible for a stance utterance to express
only pure subjectivity. It is true that there are stance utterances which overtly po-
sition the speaking subject without explicitly including any reference to a stance
object. In what appear to be simple one-place predicates (Thompson and Hopper
2001), speaking subjects position themselves subjectively – and that seems to be
the end of it. Consider examples (4) and (5), repeated here for convenience:

LANCE; I’m glad.
(33) (Hey Cutie-Pie SBC028: 52.330-53.355)
JEFF; (TSK) I’m so glad.
Up to now we have analyzed these cases as simply reflecting the speaker’s subjec-
tive self-positioning, with respect to a scale of affective value (glad). In contrasting
Lance’s I’m glad with Jeff ’s I’m so glad, what is immediately obvious as a difference
is the presence of the intensifying adverb so in the latter case. We might take this
to indicate that Jeff is positioning himself as claiming a more intense subjective
experience along the scale of gladness than Lance is claiming. While this may be
true as far as it goes, to focus exclusively on the subjective side of the equation is to
154 John W. Du Bois
leave out a key variable: the object of stance. Here the stance object is, once again,
what the speaker is affectively orienting to. But can an object be part of the stance
if it is not part of the sentence? The answer may hinge on whether orientation to
an object is taken to be a necessary part of the process of constituting subjectivity.
The evidence we have seen points to a positive conclusion: displays of subjectivity
always make relevant the relation between a stance subject and a stance object.
The claim is that it is a regular feature of subjectivity to orient to an object. If
the stance object is not overtly specified within the immediate stance utterance,
participants will feel that something is missing. If subjectivity requires orientation
to an object, the full meaning of any subjective stance must remain mysterious
until we locate the object, even if this requires us to search the discourse context
to find it. This explains why two stances in which similar or identical words are
used may still differ substantially with respect to what they are stances about.
Such appears to be the case in examples (32)–(33) above. The difference between
Jeff ’s and Lance’s stances may hinge primarily on what they are glad about – the
object of stance – but this is left unmentioned within the stance utterance itself.
But that’s not to say that the participants are not orienting to a stance object. Once
we take into account the sequential context in which the stances developed in the
first place, their actual significance becomes clear. In the first I’m glad example,
Lance, an apprentice air traffic controller, is being debriefed after a work session
by Randy, his trainer:

1 RANDY; N:o significant problems.
2 (0.5)
3 #to #dea– —
4 .. to talk about.
5 (2.0)
6 LANCE; I’m glad.
Here the stance content of the utterance I’m glad emerges from successive contri-
butions by Randy and then Lance, with the cumulative result being paraphrasable
as something like I’m glad (that there are) no significant problems to talk about.15
In the second example, Jeff is talking on the phone to Jill about her friend who is
visiting her, and asks:
(35) (Hey Cutie-Pie SBC028: 49.985-53.355)

1 JEFF; Are you guys having fun?
2 JILL; Y:es:.
3 (0.6)
4 JEFF; (TSK) I’m so glad.
Quite a different stance emerges from this sequential contextualization, para-

phrasable as I’m so glad you guys are having fun.16
The main difference in meaning between the last two examples turns out to
have little to do with the presence of the intensifier so, but is instead based on
differences in the dialogic sequence through which the stance emerged. To take
into account only the current stance utterance (I’m glad) would be to miss out on
what the stance predicate glad is pointing to. People are not glad in general, but
glad that (something). Obviously it makes a difference whether the complement
of glad is to be understood as you guys are having fun or there are no significant
problems to talk about. Even self-positioning presupposes an object, namely, the
specific entity or state of affairs toward which the speaker expresses their subjec-
tive stance.
The following diagrams represent the relevant relationship between stance
subject and stance object:
(36) Stance Positions/ Stance

# Speaker Subject Evaluates Object
6 LANCE; I ’m glad {there are no significant problems}
4 JEFF; I ’m so glad {you guys are having fun}
Along the same lines, the next stance utterance appears at first to be a simple
display of amazement:
(38) (Lambada SBC002: 665.79-667.35)

69817 MILES; I’m just ama:zed.
But the sequential context makes it clear that Miles is not just amazed, he is
amazed about something:
(39) (Lambada SBC002: 660.75-667.35)

696 MILES; Cause there’re a lot of women out there who apparently
don’t believe in using condoms.
697a (1.0)
697 PETE; Hm.
698a (0.6)
698 MILES; I’m just ama:zed.
Summing across the full discourse context, Miles’ stance amounts to something
like I’m just amazed (that) there’re a lot of women out there who (apparently) don’t
believe in using condoms. Clearly, the stance act of affective self-positioning (as
156 John W. Du Bois
glad or amazed) is incomplete until we include the object of stance – what the
speaker is glad or amazed about. Although it may be merely implicit in the cur-
rent stance utterance, the stance object is an indispensable component of even
a subjective stance. Even in its absence, the stance object remains relevant and
hence may trigger a search for it in the prior discourse. The conclusion: Subjectiv-
ity takes an object.
If subjectivity must have its object, this should be as relevant to epistemic
subjectivity as to affective subjectivity. Consider examples (7)–(8), repeated here
for convenience:

KENDRA; I kno:w.
(41) (Risk SBC024: 588.565-588.970)
On the face of it, Kendra positions herself as knowing, while Dan positions him-
self as not knowing. At least this is what the words taken literally and in isolation
seem to mean. But such interpretations in isolation are meaningless. How far off
the mark they are becomes evident once we take into account the dialogic con-
text:
(42) (Appease the Monster SBC013: 919.23 -921.13)

WENDY; Those are good spatula[s].
KENDRA; [I] kno:w.
(43) (Risk SBC024: 585.630-588.970)
JENNIFER; .. Are we not attacking each other until we get rid of the
striped guy?
Kendra and Dan each speak no more than three words, yet manage to produce ap-
parently complete stance utterances thereby. But the stances they achieve through
their slender utterances are more complex than what is immediately evident in
the stance utterances themselves. Their respective stances emerge only from the
larger dialogic sequence, as shown in the following representations:
2 KENDRA; I know {those are good spatulas}

2 DAN; I don’t know {if we are not attacking each other
until we get rid of the striped guy}18
The point is that people do not normally present themselves as knowing (or not
knowing) in the abstract. Rather, they know (or don’t know) particular things.
Generally the precise specification of what they know, if not present in the sen-
tence itself, is already there in the immediate prior discourse – often in the utter-
ance of a dialogic partner. It takes Wendy’s and Kendra’s utterances together for
Kendra’s stance to emerge. Likewise, it takes Jennifer and Dan working together
to articulate Dan’s emerging stance. This kind of co-action, as realized in the con-
text of conversational interaction, is a big part of what it means for linguistic ac-
tion to be dialogic.
Once our analysis systematically incorporates the dialogic co-participants’
contributions to the emerging stance, several things start to become clear. First,
a stance verb like know often points to a dialogically prior stance, articulating a
precise indexical relation to it. Second, the prior stance must be incorporated ana-
phorically into the interpretation of the overall emergent stance which culminates
in the current stance utterance. Third, the stance act is not necessarily complete
within a single intonation unit, clause, sentence, or even turn. Stance utterances
like I know and I don’t know are designed to incorporate their dialogic antecedents,
through which they gain the interpretative specificity they need to be complete.
This has important consequences for our understanding of subjectivity. Despite
initial appearances, the stance which culminates in a short and apparently simple
utterance like I’m glad or I know cannot be a matter of subjectivity in isolation,
whether affective or epistemic. Rather, it necessarily combines a subjective and an
objective component. In constituting subjectivity, the requirement for inclusion
of a stance subject (I) is intimately connected to the requirement for inclusion of
a stance object (the state of affairs that the speaker is glad about, informed about,
and so on). The link that is constituted between subject and object creates a vec-
tor of subjectivity. The subject-object link is often achieved dialogically, through
separate but coordinated contributions by several co-actors, as successive stance
utterances – stance and counterstance – are deployed in response to each other.
Despite popular conceptions of subjectivity as a purely internal, solipsistic state of
the individual psyche, we see from the evidence of stancetaking that the presence
of a subjective element in no way precludes the presence of an objective element
as well. In the end, subjectivity proves meaningful only when subject and object
are defined in relation to each other.
158 John W. Du Bois
Though we have not yet arrived at a full understanding of what is needed

to contextualize and interpret stance, we have come far enough to draw some
preliminary conclusions. The three questions about stance contextualization that
were introduced at the outset – regarding the identity of the stancetaker, the ob-
ject of stance, and the prior stance being responded to – make for a good start
toward filling in the information that we need to interpret any given stance utter-
ance. But the full significance of these bits of information must remain obscure
until we understand better why we should need them in the first place. We need
to ask what else may be needed to complete the picture of the stance in ques-
tion. So far, we have identified a regular pattern whereby a single stance utterance
culminates in the bringing together of an object-centered act of evaluation and
a subject-centered act of positioning. This predictable pattern remains relevant
even if one of the elements is not explicitly verbalized within the current stance
utterance, but is to be located anaphorically in the discourse of a prior speaker.
It is in fact routine to find complex stance culminations, precipitated in the end
by a seemingly simplex stance utterance, but dependent on the prior context. In
many such cases, it is possible to distinguish the object-centered element from the
subject-centered element, and to attribute each element to a distinct portion of
the discourse, whether as a specific word or phrase in the current or prior stance
utterance. The concrete localization of the overt words or other meaningful ele-
ments which ground the various components of stance is important for assessing
the compositional contribution of the several evaluative words and constructions
in a stance utterance. In the end, however, no stance stands alone. It is the stance
utterance with its dialogic context that is the relevant unit for stance interpreta-
tion. In order to arrive at a situated interpretation for any particular stance utter-
ance, it is necessary to monitor the developmental history of the emergence of the
stance, as its content is enriched via the structure of expectations regarding the
necessary recurrent components in the general model of stance.
But still more is needed if we wish to grasp how the contextually recovered
stance components combine with the explicit components of the stance utterance
itself to yield the stance interpretation. To get to the next level of understanding,
a key issue that remains to be addressed concerns the role of intersubjectivity. As
with the subjective and objective elements of stance, the place of the intersubjec-
tive in relation to other elements of stance will prove essential in arriving at a
more complete picture of the interpretive matrix within which stance is dialogi-
cally realized.
5. Intersubjectivity and alignment
We have seen how subjectivity figures in stance, most evidently in the case of
positioning, where what is positioned is typically the speaking subject – the stan-
cetaker. A positioning utterance like I’m glad foregrounds its subjectivity via overt
cues such as first-person pronouns, affective predicates, and other elements that
index salient aspects of the speaking subject. Moreover, the subjective stance pre-
sumes an orientation to an object, whether overtly present in the stance utterance
(I don’t like those) or not (I’m amazed). But even after identifying the missing
stance object, we still only have part of the story. Some hint of what remains to be
incorporated is suggested by a simple observation. In the dialogic realization of
stance, the subjective orientation to a stance object may be shared among more
than one participant. Indeed this object-orientation may extend across multiple
stance acts by different speakers. This gives rise to what I call the shared stance
object. (For the related notion of joint attention, see Hobson 1993; Kidwell and
Zimmerman 2006; Moore and Dunham 1995; Tomasello 1999; Tomasello et al.
2005.) As we will see, the shared stance object becomes the cornerstone of the
dialogic construction of intersubjectivity.
What are we to make of intersubjectivity? How does it enter into the dialogic
realization and interpretation of stance? As the word itself suggests, intersubjec-
tivity presupposes subjectivity, at least etymologically. But to observe intersub-
jectivity in action, we will have to expand our view to encompass more than one
subjectivity – to bring into focus the sociocognitive relations that arise between
two subjectivities, when the subjective stances of two participants collide within
a dialogic exchange. When we learn to see how one speaker’s subjectivity reacts
to another’s subjectivity, we will have a real opportunity to witness the dialogic
emergence of intersubjectivity.
Consider the following exchange, which provides a larger window of context
for example (29):

1 SAM; I don’t like those.
2 (0.2)
3 ANGELA; I don’t either.
Through this kind of sequential juxtaposition of evaluative stances, intersubjec-

tivity rises to focal prominence. Analogical relations are established between the
juxtaposed stances (I don’t like those : I don’t either). The foregrounding of this
dialogic relation potentially invites inferences based on the comparison. A similar
160 John W. Du Bois
dialogic relation arises between the two participants’ stances in the next example
as well:
(47) (A Tree’s Life SBC007: 581.32-585.99)
1 ALICE; I don’t know if she’d do it.
2 (0.6)
3 MARY; I don’t know if she would either.
It is useful to have a way to represent the implicit structure of the stance parallels
in such cases, in a way that perspicuously captures the potential for participant
inferencing about stance. One tool that has proven useful for representing dia-
logic relations between stances is what I have called the diagraph (Du Bois 2001).
While the diagraph was developed for independent purposes – primarily for ana-
lyzing the structural relations that characterize dialogic syntax – it can be equally
effective in bringing out the similarities and contrasts between stances in dialogic
sequence. Foregrounding the relevant structural relations, diagraphs display the
parallelism of elements using a representation like the following (abstracted from
example 47):
(48) 1 ALICE: I do n’t know if she ’d do it .

3 MARY: I do n’t know if she would either.
A word is in order regarding the organization and significance of diagraph analy-

sis. The diagraph is designed to represent the mapping of structured resonances
across utterances. (Diagraph, from dia- ‘across’ + graph ‘mapping’, means liter-
ally ‘mapping across’.) The two most immediately relevant dimensions of the
diagraph are, in simplistic terms: rows correspond to utterances,19 and columns
correspond to the specific elements which resonate between different utterances,
which are specified as parallel within the diagraph. To display the resonance icon-
ically, the diagraph is formatted so that elements that resonate with each other
are aligned vertically in columns. A further element that is always included in the
diagraph is speaker labels, which serve to index the identity of the participants
who enact the dialogic resonance in question. The recognition of a central role for
participant identity in resonance relations is part of what makes dialogic syntax
dialogic. Note that the diagraph is not intended to capture everything about the
utterances represented, but is simply one tool to be used in conjunction with oth-
ers. The diagraph focuses primarily on those elements which display significant
resonance, foregrounding their relationships to each other. As such, the diagraph
representation is not supposed to replace the original transcription with all its de-
tail, but rather should remain implicitly linked to it for purposes of interpretation.
The connection between the two representations of the same discourse excerpt is
made explicit via such notational devices as line numbers which cross-reference
each line of the diagraph with the corresponding line of the transcription. (For
more detailed discussion and exemplification of diagraph analysis, see Du Bois
2001. For related work on structural parallelism in discourse, see Blanche-Ben-
veniste et al. 1991; Harris 1952; Jakobson 1966, 1981; Johnstone 1994; Linell 1998;
Schegloff 1996; Silverstein 1984; Tannen 1987.)
Now, what does the diagraph in (48) tell us? One speaker takes a stance, then
a second speaker takes a seemingly equivalent stance. But a closer examination
of the diagraph makes it clear that, along with their parallels, the two stances are
also fundamentally different. Though subtle, the differences turn out to be impor-
tant because of what they tell us about the dialogic relations that are established
through the sequential realization of stances. The diagraph in (48) shows how
Alice responds to Mary’s I don’t know if she’d do it with a very similar utterance: I
don’t know if she would either. As similar as these two stance utterances are, there
is a limit to their convergence. The second stance utterance ends with the word
either, and this is no mere adornment. If Mary had responded to Alice’s utterance
with a lexically identical utterance – just I don’t know if she’d do it – the effect
would likely be perceived as somewhat strange, in part because of the absence of
the word either.20 The strangeness cannot be explained away as a problem with an
“echoic” utterance: saying just I don’t know if she would would be pragmatically
aberrant as well, in more or less the same way. Why is the presence of the word ei-
ther so crucial here? The same issue arises in I don’t either in (46), suggesting that
there is a general principle involved. In both cases, the word either in this con-
struction serves to index a specific intersubjective relation between two speakers
engaged in dialogic interaction. While space precludes full exploration of the de-
tailed workings of this pattern here, the evidence from many similar cases makes
it clear that either cannot normally be omitted from the second stance utterance
without causing pragmatic anomaly (Du Bois 2004). There is no other explana-
tion for the virtually obligatory presence of either in such sentences than its role
in indexing the intersubjective relationship between two stances in dialogic juxta-
position. In general terms, whenever an interactionally salient dialogic resonance
arises between two stances, the intersubjective relationship between one’s own
resonating stance and that of the prior speaker must ordinarily be acknowledged
indexically – if one wishes to avoid being judged interactionally incompetent.
One way to look at stance is to ask how it is constituted as an action within
an interpretive framing erected by the ongoing dialogic activity. In this light, the
diagraph in (48) can be interpreted in terms of who leads and who follows. The
first stance (Alice’s in line 1) can be characterized as a stance lead, while the sec-
ond (Mary’s in line 3) is a stance follow.21 The importance of this contrast becomes
clear when we note that, despite their similar stance content, the participants
mark their stances differently. Mary’s stance follow is virtually required to include
162 John W. Du Bois
the word either, which serves an intersubjective indexical function here. In con-
trast, Alice’s stance lead is virtually precluded, pragmatically if not grammatically,
from including the intersubjective indexical either. Example (46) is like (47) in
this respect: the intersubjective use of either is pragmatically required. The subjec-
tive positioning in (46) is affective (don’t like), while the subjective positioning in
(47) is epistemic (don’t know), as well as modal (if) in effect. But the need to index
the intersubjective relationship between a stance follow and a prior stance lead
remains the same.
There is an important connection between intersubjectivity and the stance
act of alignment that is visible in these examples. Note that alignment, as I use the
term, represents a point along a continuous scale or range of values. In contrast to
common usage which forces a binary choice between a positive pole (referred to
as aligned) vs. a negative pole (disaligned), the approach I favor treats alignment
as continuously variable in principle. By recognizing the variability of scalar align-
ment we can take into account the fact that stances are aligned by subtle degrees,
so that stance alignment can be relatively positive or negative – or, more precisely
speaking, convergent or divergent to some degree. Alignment is in play whether
the direction is convergent, divergent, or as often happens, ambiguous between
the two. Thus two participants in dialogic interaction should be understood as
engaging in the alignment process when they converge to varying degrees, and,
by the same token, when they diverge to varying degrees. In this light, the use
of either in the cited examples should be seen as part of an act of alignment that
serves to calibrate the intersubjective relationship implicit in the stances of en-
gaged co-participants. In such cases, words like either (or too, as discussed below)
can be said to function as intersubjective alignment markers.
6. The stance triangle
We have been assembling an analytic toolkit of interconnected concepts and

methods designed to shed light on the various elements and processes of stance.
Key components include the concepts of evaluation, positioning, and alignment,
as well as the sociocognitive relations of objective, subjective, and intersubjective
intentionality. In developing new ways of analyzing these and other elements, we
have been working to lay the groundwork for the next stage of theorizing stance.
Now it is time to bring all the elements together to forge a unified framework for
stance. The picture we are moving toward is one in which stance is seen as a single
unified act encompassing several triplet sets of distinct components and process-
es. On the level of action, stance is to be understood as three acts in one – a triune
act, or tri-act. To the question posed at the outset (in Section 2) as to whether
evaluation, positioning, and alignment represent three different types of stance,

the view from the stance triangle suggests that they are simply different aspects of
a single stance act. Rather than three separate types of stance, we interpret them
as subsidiary acts of a single overarching, unified stance act. Each subsidiary act is
distinguishable from the others by virtue of its own distinctive consequences, yet
the three are yoked together through their integration in the dialogic stance act.
The stance act thus creates three kinds of stance consequences at once. In tak-
ing a stance, the stancetaker (1) evaluates an object, (2) positions a subject (usually
the self), and (3) aligns with other subjects. The following definition sums it up:
(49) Stance is a public act by a social actor, achieved dialogically through overt
communicative means, of simultaneously evaluating objects, positioning
subjects (self and others), and aligning with other subjects, with respect to
any salient dimension of the sociocultural field.
Alternatively, adopting the first-person point of view of the stancetaker as speak-

ing subject, we can informally gloss this definition as follows:
(50) I evaluate something, and thereby position myself, and thereby align with you.
Figure 1. The stance triangle

164 John W. Du Bois
The beginnings of a framework for analyzing stance are implicit in these defini-
tions. In the rest of this section, I will further sketch out a preliminary version of
this framework, along with some initial suggestions as to how it can expand the
analytic reach of the stance concept.
The clearest way to represent the stance model I am proposing is in the form
of a triangle (Figure 1).
The three nodes of the stance triangle represent the three key entities in the
stance act, namely the first subject, the second subject, and the (shared) stance
object. The three sides of the triangle represent vectors of directed action that
organize the stance relations among these entities. While the stance triangle com-
prises the three subsidiary acts of evaluating, positioning, and aligning, these are
not distributed evenly among the three sides, as in the expected one-to-one cor-
respondence found in conventional triangular models. Rather, two of the three
sides represent evaluative vectors directed from one of the two stance subjects
toward the single shared stance object. The first evaluative vector originates from
the first subject, the second from the second subject. The third side of the triangle
(the vertical line on the left) represents alignment between the two subjects. Sig-
nificantly, each of the three stance act vectors is relational and directed, linking
two nodes of the triangle. Vectors of alignment may originate in either the first
or second subject and be directed toward the other subject. For each vector of
directed action in the diagram, an arrowhead points in the direction of action’s
object or target. Because there are two social actors represented in the stance tri-
angle – the first and second stance subjects – there are two tokens of each action
vector type. This makes for a total of six arrowheads, corresponding to the three
acts of evaluation, positioning, and alignment, doubled by the co-presence of two
subjects. The stance triangle is unusual in that it depicts three stance acts for the
first subject, and again the very same three stance acts for the second subject.22
And yet the acts are different, the second time around.
The stance triangle provides the basis for understanding the causal and infer-
ential linkage that may arise between the various subsidiary acts. Concomitant to
evaluating a shared stance object, stancetakers position themselves. Concomitant
to positioning themselves, stancetakers define alignment with each other, wheth-
er the alignment is convergent or divergent. Depending on the circumstances,
it is possible to draw inferences regarding any unspecified portion of the stance
triangle, as long as the rest of the triangle is known. Crucially for the analysis, all
three of the three-in-one subsidiary acts remain relevant to stance interpretation
even if only one or two of them are overtly expressed in the linguistic form of the
stance utterance. The stance triangle shows how a stance utterance that specifies
only one of the three vectors can allow participants to draw inferences about the
others. For example, if Melissa agrees with Jan, she positions herself as taking
the same stance (roughly speaking) as Jan, including the evaluation Jan has per-
formed in her own prior stance. Conversely, if Melissa expresses an evaluation
that is effectively the same as Jan’s previous evaluation, this allows us to infer her
convergent alignment with the previous speaker. In sum, the stance triangle pos-
its a model of the components of stance and of the organization of the relations
between them, with strong implications for inferencing regarding participants’
positioning, alignment, and evaluation. Attending to the structured interrelations
among the acts and entities which comprise stance allows participants, and ana-
lysts, to draw inferences by triangulating from the explicit components of stance
to the implicit. What I am proposing is that the structure of dialogic action repre-
sented in the stance triangle offers a framework for analyzing the realization and
interpretation of stance. Our understanding of stance is enhanced, I suggest, by
taking seriously the interrelations among components of the stance act as speci-
fied in the stance triangle model.
To assess these claims for the theoretical significance of the stance triangle,
we need to see it in action – to test its utility in the analysis of actual instances of
stance in interaction. We begin by taking a second look at an example introduced
earlier (46), considering it now in light of the stance triangle:

1 SAM; I don’t like those.
2 (0.2)
3 ANGELA; I don’t either.
The three entities at the nodes of the stance triangle are more or less transpar-
ently represented in this example – the first stance subject (Sam’s I), the second
stance subject (Angela’s I), and the shared stance object (in Sam’s utterance, those;
in Angela’s, what some would call a zero, or a deletion, representing the under-
standing that Angela is referring implicitly to the same referent as Sam’s those).
Sam’s stance predicate (don’t) like serves both to position the entity expressed by
its syntactic subject (I) and to evaluate the entity expressed by its syntactic object
(those). While the stance acts of evaluation and positioning are more or less evi-
dent from direct inspection of the conversational example, to see the alignment
clearly it will be useful to create a diagraph:
(52) 1 SAM; I do n’t like those .

3 ANGELA; I do n’t either.
Angela marks her contribution as a stance follow to Sam’s stance lead, deploying
the word either in its intersubjective alignment function.
166 John W. Du Bois
To display the analysis in terms of the stance triangle more precisely, we can
incorporate labels specifying which entities and actions are present in the stance
utterance, and how they are expressed (or implied) in it, as in the following dia-
gram, termed a stance diagraph.23

# Speaker Subject Evaluates Object Aligns
1 SAM; I1 don’t like those
3 ANGELA; I2 don’t {like} {those} either
As for the three stance actions, in these data, the verb specifies both the evaluation
of the object and the positioning of the subject, so the two labels are combined in
a single column. Angela’s use of the word either indexes alignment, taking account
of the fact that Angela’s stance utterance is a stance follow which builds dialogi-
cally off of Sam’s prior stance lead. The column heading thus marks the alignment
function of the word either accordingly. Note that this stance diagraph representa-
tion is only intended as an informal aid to visualizing those elements in the stance
utterance which correspond most directly to the relevant stance triangle entities
and actions. For example, if we really want to specify how alignment is achieved
in this kind of utterance, we would have to acknowledge that it’s not just using the
word either that does it, but also the resonance generated through the act of repro-
ducing words and structures of the prior speaker. There is always more to stance
in dialogic interaction than can be captured in any labeled diagram as simple as
this one. Nevertheless, the stance diagraph helps to make visible key aspects of
the mapping between forms which resonate across utterances. This brings out
the similarities, but just as important, the differences that constitute what I have
called the stance differential. The stance diagraph serves as a useful intermediate
stage in the analysis leading to the stance triangle, especially in foregrounding the
stance act of alignment. Having parsed out the various component acts and enti-
ties of a stance exchange via the stance diagraph, one can in principle then map
this analysis onto the stance triangle.
The next example provides a further test case involving the three subsidiary
acts of evaluation, positioning, and alignment, as well as the stance differential.
(The first line of this example was analyzed above in (17) and (24).) As is clear
from the prior discourse, Ken is talking specifically about going to Nicaragua.
Now look at what develops next, in the subsequent exchange of stances:

1 KEN; .. I would love to go:.
2 (0.3)
3 LENORE; Yeah.
4 (1.0)
5 JOANNE; Yeah?
6 (0.9)
7 I want to go too.
There are interesting aspects of resonance to be found in various of the possible

pairings of stance utterances that could be made here, but two resonance pair-
ings in particular are of immediate interest: the two yeah’s (lines 3 and 5), and I
would love to go : I want to go too (lines 1 and 7). First, consider Lenore’s yeah in
line 3. Uttered with a final falling intonation, this amounts to an intersubjective
alignment marker, displaying her convergence with Ken’s stance in line 1, as the
relevant prior stance. After a one-second pause – a saliently long pause for a lively
conversation like this – Joanne utters the same word as Lenore, but with a differ-
ent intonation. Even though the words are identical, a stance differential is created
nonetheless, primarily by intonation but also by other factors such as voice quality
and sequential placement. Where Lenore had a terminative intonation contour
(realized as a final fall) on Yeah, Joanne uses a rising appeal contour, suggesting a
questioning of, or at least an ambivalence toward, Lenore’s more definite yeah.24
Joanne’s display of a subtle stance differential in line 5 effectively raises some
question as to whether she will fully align herself with the prior salient stance, to
which Ken and Lenore are both by now on-record adherents. After another sub-
stantial pause, Joanne goes on in line 7 to commit to a sort of stance follow. But
she withholds the word love, choosing want instead, a presumable downgrade of
the level of commitment to the proposition at issue.25 This sheds light on her lack
of commitment as implied by her use of a questioning intonation for what could
have been, with different intonation, a strong marker of convergent alignment.
The crucial resonance relations in this exchange are highlighted in the following
diagraph of Ken and Joanne’s utterances:
(55) 1 KEN; I would love to go .

7 JOANNE; I want to go too.
While not realized in immediately successive utterances, the analogy between the
two stance utterances (I would love to go : I want to go too) is unmistakable – suf-
ficiently transparent to be recognizable across a few intervening utterances. Al-
though it might appear that Joanne is saying more or less “the same thing” as Ken,
the differences which constitute the stance differential are actually quite crucial,
as the diagraph above suggests. To create a more explicit representation labeling
the relevant stance entities and actions, we recast (55) as a stance diagraph:
168 John W. Du Bois

# Speaker Subject Evaluates Object Aligns
1 KEN; I1 would love to go {to Nicaragua}
7 JOANNE; I2 want to go {to Nicaragua} too
This diagram shows clearly how Ken performs a stance lead, and Joanne a stance
follow. Joanne explicitly registers the difference between her follow and his lead via
the word too, which is not omissible in this context without pragmatic anomaly.
The evidence from many such cases makes it clear that when a stance follow
is juxtaposed to a dialogically resonant prior stance lead, then, given certain well-
defined conditions,26 the intersubjectivity involved must be registered obligato-
rily, using the appropriate linguistic index: too for positive utterances, either for
negative utterances.27 Although the role of intersubjectivity in language usually
remains implicit or is subtly expressed (e.g., via prosody or sequential placement),
sometimes it is realized overtly in the tangible form of a specific word, as in this
case. As this and earlier examples attest, the indexical function of too and either
becomes a valuable diagnostic for the individual speaker’s obligatory engagement
with dialogically constructed intersubjectivity.
One further point about the stance triangle calls for comment here. In the
present analysis of stance, the shared stance object obviously plays a critical role,
binding the subjectivities of dialogic co-actors together, thereby articulating an
intersubjective relation between them. But what about cases which don’t seem to
involve a shared stance object? This would appear to present a challenge for the
present analysis. Yet the argument can be made that the stance triangle applies
even in such less-than-transparent cases. While this question deserves a more
extended response than can be presented here, a word of commentary may give
some idea of what kind of answer will be required. It has been claimed that all
meaningful use of human language, from the age of about one year, presupposes
shared orientations, for example toward a word’s referent (Tomasello 1999; To-
masello et al. 2005; see also Hobson 1993). It will be an important task for future
research to show how the stance triangle extends naturally to incorporate such
observations. In the meantime, in cases where it may not be obvious that the full
stance triangle is in play, it is usually possible to break the triangle down into its
component vectors (e.g., an individual stance vector constituting the subject-ob-
ject evaluative relation), and thereby to achieve an insightful, if partial, analysis
of stance.
At the outset (Section 2) we considered whether we could set up a distinction
between different types of stance. Among the more promising candidates were
evaluation, positioning, and alignment. But should we consider these as distinct
types, or merely different facets, of stance? The answer may hinge on whether
they regularly occur separately, in complementary distribution, or whether they

rather co-occur as different aspects of a single act of stancetaking. The evidence
we have seen suggests that they are not so much alternatives as different facets of
a more general phenomenon of stance. This is what the stance triangle tries to
represent. Since its first introduction (Du Bois 2002a, 2002b) the stance triangle
has come to be used in a diverse and informative body of work on stance in in-
teraction (e.g., Haddington 2004, 2005, 2006; Kärkkäinen 2006; Takanashi 2004;
see also the papers by Kärkkäinen, Haddington, Keisanen, and Rauniomaa in this
volume). The stance triangle has proved well suited to framing effective research
questions, without imposing predetermined limitations on what the researcher
can undertake to investigate about stance.
7. Discussion
The conception of stance we need is one capable of being situated within a larg-
er mediating framework for linguistic action and interpretation, which is itself
grounded in the dialogic dimensions of sociocognitive relations, interactional
collaboration, and shared responsibility between conversational co-participants.
To come to terms with the complexity of the stance act and its interpretive frame,
it is necessary to articulate a systematic approach to understanding what stance is.
To do this, we need a way to represent how stance works. We need this if we want
to be able to frame questions about how discourse participants achieve stance,
and about how the multiplex meanings and consequences of stance play out in
the public space of interaction. The framework I have proposed is encapsulated in
a triangle, but it must not be forgotten that there is a larger interpretive apparatus
that subtends this geometric metaphor. The stance triangle is built on certain ba-
sic assumptions about what is needed to constitute a stance, which I have elabo-
rated in some detail in this paper.
I have argued that stance can be analyzed, in its fundamental structure, as
a single unified act encompassing three subsidiary acts – in effect, a triune act,
or tri-act. I define stance as a public act by a social actor, achieved dialogical-
ly through overt communicative means (language, gesture, and other symbolic
forms), through which social actors simultaneously evaluate objects, position
subjects (themselves and others), and align with other subjects, with respect to
any salient dimension of value in the sociocultural field. I have argued for a par-
ticular configuration of actors and actions as the defining feature of stance. Key
to this configuration is a set of three entities (first subject, second subject, stance
object) and a set of three actions (evaluation, positioning, alignment). The analy-
sis of stance in terms of these elements lays the basic foundations on which the
170 John W. Du Bois
stance triangle is built. But there is more to the stance triangle than this. Crucially,
the stance triangle specifies the effective relationships between entities and acts
through vectors of stance action. On another level, sociocognitive relations link
the subject and object of stance relations, and map one subject-object vector to
another to constitute the intersubjective relation. The model shows how stance
can be analyzed in terms of a set of triangular relations which link entities via
vectors of dialogic co-action and intersubjectivity.
What consequences flow from the triangular model of stance? Here I will
point to just a few of the implications that arise. The parallelism of two of the three
stance vectors allows us to analyze the phenomenon of alignment, based on an
analogy which is partly structural and partly functional. This parallelism may in
turn invite analogical inferences (Itkonen 2005; Jakobson 1966; Silverstein 1984).
The stance triangle provides a framework for understanding the sociocognitive
relations (objective, subjective, intersubjective) that are present in all dialogic
interaction, and tries to clarify how these relations are constituted through the
stance acts of evaluating objects (objective), positioning subjects (subjective), and
aligning with other subjects (intersubjective). Especially in cases involving what
I have called the stance differential, the triangle provides leverage to analyze the
fine calibration of convergent and divergent alignment in the stancetakers’ posi-
tioning of themselves relative to others.28
In the world of theory-making, there are many triangles. Why one more?
But this is not your average triangle. Most triangle diagrams try to maximize the
contrast between three terms selected to display the most important and sharply
differentiated abstract concepts in the theory. Here, on the other hand, two of the
three points of the triangle are used to represent what amounts to the same thing
twice – two stance subjects. This is not done to flout the principle of economy,
but because it displays the interchangeability of perspectives (Mead 1934; Schutz
1962) of dialogic stance partners who alternately fill the roles of speaker and hear-
er, of first subject and second subject, as the polarity of co-action cycles rapidly via
the dynamics of dialogic exchange.
Unlike many triangles which remain in the realm of the theoretical, this tri-
angle is meant to be used. The idea is that its general architecture underlies the ac-
tual practices of realizing stances and negotiating their significance in particular
events of language use. To the extent that it articulates a predictable framework for
action, the stance triangle is available to participants as a resource for organizing
their evaluative actions on any specific occasion. Looked at from the analyst’s per-
spective, this makes it relevant to describing what is happening in actual instances
of stance in interaction. Appropriately deployed, the stance triangle can clarify the
array of entities and sociocognitive relations that are activated, constituted, and
brought into relation to each other by a particular stance action. In addition, by
allowing us to triangulate between one participant’s observable (if partial) stance

actions and those of their dialogic partner, the stance triangle provides a way to
apprehend the regularity of stance inferencing that brings systematicity to the
consequences of stance.
8. Conclusions
Stance is not something you have, not a property of interior psyche, but something
you do – something you take. Taking a stance cannot be reduced to a matter of
private opinion or attitude. Using the language of Wittgenstein (1953) we might
say: There are no private stances. We deploy overt communicative means – speech,
gesture, and other forms of symbolic action – to arrive at a dialogic achievement
of stance in the public arena. Stance can be imagined as a kind of language game
in Wittgenstein’s sense, which is to say, it unfolds within a recognized framework
for interpreting action. To realize stance dialogically means to invoke a shared
framework for co-action with others. Stance must be intelligible within that
framework, as it comes into existence in its natural environment of dialogic in-
teraction. The expectation of a well-defined framework for stance remains in play
even if the framework must in part be dynamically constituted by the participants
themselves in the very act of taking a stance. Stance is an activity built for two (or
more). As we maneuver within the constantly shifting field of stances, we find
that even our own stance must be enacted collaboratively. Through joint and sev-
eral acts we engage in the activity of stance, invoking relevant components of the
stance frame as we both shape and respond to the multiplex consequences which
flow from our actions. Our exceptional agility at managing the dialogic play of
stance and counterstance is underlain by an implicit awareness of the structure
of the activity system that frames and enables the achievement of stance. Par-
ticipants use their knowledge of the elements, actions, and vectors of stance – as
described here by the stance triangle – to project the multiplex consequences of
their own and their partners’ unfolding stances.
Stance is best understood in terms of the general structure of the evaluative,
positioning, and aligning processes that organize the enactment of stance, rather
than as a catalog of the contents of stance, or even – as important as these may
be – of the sociocultural value categories that are referenced by stance. This is what
the stance triangle tries to achieve: a general level of analysis that can be applied in
principle to any instance of stance.29 Stance always combines elements of general-
ity and specificity; but while the stance triangle as a theoretical object is general
by design, it is also intended to frame the concrete analysis of specific stances. The
general principles governing the stance framework, including principles of dialogic
172 John W. Du Bois
co-action and the intersubjective organization of sociocognitive relations, impinge

on the realization and interpretation of individual stances. Part of the meaning of
any stance derives from the manner of its realization in interaction, including the
dialogic dimensions of the form it takes. Stance utterances gain added levels of
significance through their juxtaposition with other stance utterances. Much of the
dialogic quality of stance comes from the way a present stance may resonate with
a prior stance, as resonance across stances shapes the sociocognitive alignment
between speakers, and thus helps frame intersubjectivity.
The method used here to analyze resonance across successive stance utteranc-
es is dialogic syntax (Du Bois 2001), and specifically the diagraph. The diagraph
representation depicts the precise resonance relations between two stance utter-
ances, allowing us to probe in finer detail the specific reciprocal relations between
sub-portions of the utterances mapped in the diagraph. Given an explicit set of
mapping relations between counterpart elements across paired stance utterances,
we can go on to draw detailed inferences regarding aspects of the stance differ-
ential between the two participants, including convergences and divergences of
epistemology, affect, and other regular concerns of stancetakers. The potential is
extremely rich for analytical and theoretical cross-fertilization between the meth-
ods of the stance triangle and the diagraph. This will be fruitful as long as we
remember that dialogic processes must be real for participants if they are to found
the claims about stance which are implicit in the stance triangle.
From a dialogic perspective, no stance stands alone. Each stance is already
specific with respect to, for example, the participants it indexes, the objects it
evaluates, and the dimensions of sociocultural value it invokes. But beyond this,
any stance realization is capable of extending its particularity further by entering
into unique relationships with other stance realizations, within particular con-
figurations of dialogic exchange. This sets up dialogic juxtapositions of stances
which may foreground the structural analogy between pairs of successive stances,
yielding a combinatorial explosion of resonance relations, as these are activated
across the paired stance utterances. Yet the resources available to participants for
interpreting such a diversity of stances are sufficient to the task, while remain-
ing constant and general. The stance triangle – including its component vec-
tors – provides a general account of the framing processes which adapt to both
the particularities of the individual stance act and its unique configuration of dia-
logic-sequential development. The tension between generality and particularity,
characteristic of all culturally framed social practice, is necessarily reflected in
the interpretation and realization of stance. To resolve this tension, or simply to
embrace it, constitutes an ongoing challenge for any theory of stance, including
the present one.
Stance is undeniably complex. There is no avoiding the rapid ramification of

meaning and consequence as stances emerge from the flux of dialogic interac-
tion. Still, we might wish for something simpler – a take-home message that we
can reference to keep a clear focus on the central meaning of stance. What is it
that sets stance apart as a research agenda, a way of conceiving and analyzing the
nature of linguistic and social action? What is the ultimate import of stance? The
short answer: Stance is an act of evaluation owned by a social actor. This charac-
terization, while by no means complete, unites three key aspects of social life:
act, responsibility, and value. The act of stance is performed in the public space
of dialogic interaction, where it inherits the peculiar contingencies of all dialogic
action, both influencing and being influenced by the co-actions of others. Despite
its apparent dependency as a local symbolic action delimited by the uptake of
other individuals, stance may be among the most broadly consequential of social
actions in its cumulative effects – an act whose force can be measured by its ef-
fective penetration into virtually all domains of sociocultural life. Responsibility,
the second element in this capsule cameo of stance, comes with ownership. In
the dialogic shop of stances, there’s a rule: If you take it, you own it. Responsibil-
ity for the stance act is serious business, with potentially profound consequences
for the relationships of stance actors with their dialogically co-responsible part-
ners and with their expanding networks of social relations along wider horizons
of time and space. Because stance responsibility is of interest to all within the
stancetaking community of discourse, the question of who took which stance is
perennially salient, is remembered over time, and counts as negotiable coin in the
currency of reported discourse. Through its open-ended social circulation, stance
(and meta-stance narratives) can only expand its role in the broader calculus of
social meaning. Third, value is what stance is all about – literally. Stance always
invokes, explicitly or implicitly, presupposed systems of sociocultural value, while
at the same time contributing to the enactment and reproduction of those sys-
tems. Social actors are accountable for how they manage and indeed reshape the
systems of social value on which we all depend. In sum, ownership of stance is the
glue that binds the stance act together with actor responsibility and sociocultural
value, so that all is linked to a social actor with a name, a history, an identity. As
players in the language-game of stance, we’ve all got some skin in the game. We
make it our business to know where the other players stand, who they stand with,
and where they’re headed. And we care about the state of the game, too: how
it is played, who plays it well and fairly, in what condition the players leave the
turf – and what all of this implies for the environs of sociocultural value in which
we all must live.
We are left with a view of the stance act as, perhaps, the smallest unit of so-
cial action. Such a fundamental status will be justified if stance can be shown to
174 John W. Du Bois
bind together the minimum structures necessary to attain the force of social ac-
tion. This is what the stance triangle aims to represent: the minimum structure
of stance as dialogic action. In depicting the co-participants’ joint evaluative ori-
entation to a shared stance object, the stance triangle proposes a framework for
understanding the dialogic realization of intersubjectivity in a way that is capable
of embracing both convergence and contestation. To be sure, it may seem coun-
terintuitive to locate contestation within intersubjectivity, rather than to simply
demand mute agreement to the normative assumptions of social reality as the
conventional prerequisite to communication. But the evidence from stance in in-
teraction is clear: convergence and divergence of evaluative alignment are equally
at home in the dialogic engagement of co-participants. If the stance triangle is to
have analytic value in the end, it must come from striving to represent at once the
unity, and the ambivalence, of stance as it emerges in dialogic interaction.
Notes
1. My understanding of stance as presented here has developed during the course of a series
of presentations at conferences and symposia, including at the meetings of the American An-
thropological Association, New Orleans (2002); Volkswagen Foundation conference on Rheto-
ric Culture, Johannes Gutenberg University, Mainz (2002); Kroeber Anthropological Society,
Berkeley (2003); a lecture series on stance and dialogic syntax, Oulu, Finland (2003); LangNet
Symposium, Oulu (2003); Fulbright lecture, Södertörn University, Södertörn, Sweden (2004);
and, finally, culminating in a presentation on “The Intersubjectivity of Interaction” at the 10th
Rice Symposium, on Stancetaking in Discourse, at Rice University, Houston (2004). My re-
search on ‘too’ and ‘either’ in the Santa Barbara Corpus was presented at the Rice Symposium
and in fuller form at the International Computer Archive of Modern and Medieval English
(ICAME), University of Verona (Du Bois 2004). In addition, early versions of my ideas about
stance were presented in a series of colloquia during this period at the Linguistics Department
and the group on Language, Interaction, and Social Organization (LISO) at the University of
California, Santa Barbara. I have benefited greatly from discussion with all these audiences,
for which I am grateful. I am especially thankful for comments on earlier versions of this work
from Mira Ariel, Mary Bucholtz, Patricia Clancy, Robert Englebretson, Rachel Giora, Pentti
Haddington, William Hanks, John Haviland, Susan Hunston, Cornelia Ilie, Adam Jaworski,
Elise Kärkkäinen, Tiina Keisanen, Amy Kyratzis, Gene Lerner, John Lucy, Mirka Rauniomaa,
Geoffrey Raymond, Joanne Scheibman, Robin Shoaps, Michael Silverstein, Hiroko Takanashi,
Sandra Thompson, and an anonymous reviewer for the present volume.
2. See the Appendix for transcription conventions. Note that some of the transcriptions have
been simplified for the sake of clarity. For example, square brackets for overlapping speech have
been omitted when the cited example doesn’t include the other half of the pair of overlapping
utterances. For full transcription details, the original sources may be consulted, as described in
the following note.
3. Most of the examples in this paper are taken from conversations in the Santa Barbara Cor-
pus of Spoken American English (Parts 1 and 2). The source citation gives the title of the dis-
course in italics (e.g., Lambada), followed by the identifying number of the discourse in the
Santa Barbara Corpus of Spoken American English (e.g., SBC002). The final portion of the
citation for each excerpt consists of two numbers, representing respectively the start time and
end time in seconds. With this information it is possible for interested readers to listen to the
appropriate portion of the relevant audio file (e.g., SBC002.WAV) by accessing the Santa Bar-
bara Corpus as published on CD and DVD (Du Bois et al. 2000; Du Bois et al. 2003), or on the
internet. Further information on contents of and access to the Santa Barbara Corpus of Spoken
American English is available at http://www.linguistics.ucsb.edu/research/sbcorpus.html.
4. The predicate amazed can be understood as incorporating, in addition to its affective di-
mension, a salient epistemic dimension as well.
5. The source of this example, cited as LSAC, is the Longman Spoken American Corpus. The
LSAC is a five million word corpus of spoken American English conversation, recorded under
my direction by researchers from UC Santa Barbara. Conversations in the LSAC were recorded
all over the United States using expertise and methodology developed for the Santa Barbara
Corpus of Spoken American English and corpus design concepts developed by Longman in
their work helping to build the British National Corpus. The LSAC was developed under con-
tract to Longman publishers.
6. The names of speakers are arbitrarily assigned pseudonyms. The following is a key to the
original labels in the LSAC corpus for the speakers cited here: in LSAC 1396-01, CORA=<1828>
and LESLIE=<1829>; in LSAC 1296-02, KIM=<1564> and PAT=<1565>.
7. Alternatively, agreement can be articulated in relation to the proposition expressed by a co-
participant – the stance content of a prior utterance, as denoted, for example, by the word that
in I agree with that.
8. Names used for speakers are pseudonyms.
9. In explicating the interpretation of a stance in its dialogic context, I have sometimes ven-
tured a paraphrase of the “content” of the stance as it has emerged from the successive con-
tributions of several participants. The problem is that the paraphrase usually comes across as
too explicit or too literal in character. A summary paraphrase cannot do justice to the specific
quality of the actual stance as it emerges dialogically in discourse. This is because to paraphrase
is to attempt to give a monologic approximation of a stance that may have taken two people
to create dialogically in the first place. The challenge of precisely and perspicuously describing
stances that emerge in dialogic interaction is a serious one, which deserves more attention than
we can give it here.
10. The content of Kim’s stance is constituted as no more or no less than any other school. (This
phrasing is admittedly rather opaque in itself, but unfortunately this cannot be clarified without
expending a lot more time and space than is warranted for the present example; nevertheless,
the general point about alignment should be sufficiently clear.) This is to be distinguished from
the final portion of the utterance (is the way I see it), which pertains narrowly to Kim’s framing
of the proposition. This is a parenthetical framing of self-positioning, not part of the stance
proposition as such, so the framing phrase is interpreted as outside the scope of Pat’s agreeing
move. Alignment markers like the verb agree typically pick out just the stance content or stance
proposition, leaving the previous speaker’s metalinguistic framing or self-positioning out of the
calculus of stance alignment. How this works in detail on a structural level is a rich topic for
further analysis.
11. To be sure, Melissa’s use of the modal should introduces a subtle shift which slyly mitigates
the directive force of Jan’s original unmitigated imperative.
176 John W. Du Bois
12. For a rich treatment of agreement from the perspective of conversation analysis, see Heri-
tage (2002) and Heritage and Raymond (2005).
13. In most of the examples presented in this paper, the stance subject can be taken to be more
or less equivalent to the speaking subject. But things are not always so simple, as is clear from
the sophisticated analyses of the sujet de l’énonciation by the cited authors, as well as from char-
acterization of the multivocality of certain kinds of utterances (Bakhtin [1934] 1981; Goffman
1981; cf. Agha 2005). While the complex case of multivocality is very interesting to consider
from the perspective of the stance triangle, this must await a separate treatment.
14. In other words, they are intentional predicates, in the philosophical sense (Searle 1983).
15. There are additional interesting issues here having to do with the incremental realization
of Randy’s turn, the timing of pauses, and what could be considered a lengthy delay on Lance’s
part in coming in with I’m glad. But these issues are largely orthogonal to the present discus-
sion.
16. How this stance develops over time and across speakers is of some interest. Jeff ’s affective
stance utterance in line 4 builds off of Jill’s prior yes in line 2, which in turn endorses Jeff ’s own
setting of the question in line 1. This can be considered an instance of other-positioning (line
1). I have developed the concept of other-positioning to account for cases in which the first
subject (speaking subject) proposes a candidate stance for the second subject (addressee), with
varying degrees of impositive force. Given that this often seems to provide the best analysis for,
e.g. ordinary questions, the phenomenon of other-questioning is surprisingly commonplace.
(For application of this concept to questioning by interviewers in television news formats, see
Haddington 2005, and Haddington this volume.)
17. This number represents the intonation unit number (or line number), counting from the
beginning of the published transcription.
18. This is not to suggest that an epistemic paraphrase tells the whole story of Dan’s I don’t
know, which may have as much (or more) to do with an act of demurral that hedges and blurs
his response to (and responsibility for) the candidate stance (other-positioning, see note 16)
that Jennifer presents him with. But that’s another story.
19. More precisely, it is intonation units (Chafe 1993; Du Bois et al. 1993) that generally define
the rows (or “strands”) of a diagraph. Based on my research (Du Bois 2001), intonation units
represent the most salient and productive unit for dialogic mapping in conversation. Where
information about the intonation unit is unavailable, the most viable alternative for identifying
the rows of a diagraph would generally be the clause.
20. One can try to imagine special circumstances in which the pronoun I is given heavy con-
trastive stress (suggesting that it is only I, but not you, who don’t know if she’d do it), but this
would be quite unusual.
21. My terms here are modeled on the notion of gaze follow(ing) (Tomasello 1999: 62–67; To-
masello et al. 2005), which is parallel to the phenomenon of stance follow in interesting ways.
For a related notion, see discussion in conversation analysis of the interactional negotiation
of turn status as epistemically “first” (=lead) or “second” (=follow) (Heritage and Raymond
2005).
22. Note that the self-positioning act defines a vector which emanates from a subject and re-
flects back on that same subject. To explicitly represent both source and target of the reflexive
vector of self-positioning would involve a circular arrow originating in the subject and reflect-
ing back on itself. Rather than display such an arrow in this simple diagram of the stance trian-
gle, we show only the head of the arrow for the self-positioning vector, as it reflects back on the
stance subject. This reflection can be seen as being triggered by a sort of “blowback” from the
subject’s act of evaluating the stance object. (Other-positioning requires a significantly more
elaborate notation, and thus is not represented in Figure 1.)
23. The stance diagraph incorporates aspects of the stance diagram (see discussion of examples
(30)–(31)), such as the labeling of stance roles (stance subject, stance object) and functions
(evaluation, positioning, alignment), combined with aspects of the diagraph, such as the inclu-
sion of multiple lines representing dialogically resonating utterances, with vertical alignment of
elements to iconically display which are resonating with which.
24. While the dialogic resonance in lines 5 and 7 between Lenore and Joanne is interesting in
that the intersubjective stance differential is subtly realized through an audible contrast located
primarily in the intonational difference (Yeah. : Yeah?), it will not be diagrammed here for rea-
sons of space.
25. The passage continues with several further developments regarding this stance negotiation,
but a fuller analysis of these events must be reserved for another occasion.
26. While it is beyond the scope of this article to lay out all the conditions governing the use
of too and either, one key factor can be mentioned here: the use of subjective intentional stance
predicates such as like, love, know, think, and want (Du Bois 2004).
27. This is not to say that too and either always function as diagnostics for intersubjectivity.
Aside from their high frequency use in marking intersubjective pragmatic relations, both words
can be used to mark referential semantic relations as well (although in spoken discourse this
objective function is much rarer than the (inter-)subjective function presented here (Du Bois
2004)). Conversely, there are many intersubjective contexts in which these particular indexical
forms do not appear. While too and either should be recognized as powerful diagnostics for
intersubjectivity when their conditions of use are applicable, the analysis of intersubjectivity in
language must draw on a wide variety of additional tools, including diagraph analysis.
28. The stance triangle bears an important relation to the systems of sociocultural value that
stances invoke and reproduce, as social actors position themselves and evaluate entities with
respect to specific values along any socially salient dimension of the sociocultural field. How
stancetaking processes both invoke and construct the associated systems of sociocultural value
is a critical issue which we can only point to in this paper, but one which represents a prime
topic for further research.
29. While the inclusion of the shared stance object – seemingly a specialized property of cer-
tain kinds of stance exchanges – might seem to limit the general applicability of the stance
triangle, from a dialogic perspective it can be argued that a shared orientation to a stance ob-
ject is a general property, not only of stance acts but of the use of language in general (see the
discussion at the end of Section 6). But that is a long story, and must be reserved for another
occasion.
References
Agha, A. 2005. “Voice, footing, enregisterment.” Journal of Linguistic Anthropology 15: 38–59.
Bakhtin, M. [1934]1981. The Dialogic Imagination: Four Essays by M.M. Bakhtin. Austin, TX:
University of Texas Press.
Benveniste, E. 1971. “Subjectivity in language.” In Problems in General Linguistics, E.M. Meek
(trans), 223–230. Coral Gables, FL: University of Miami Press.
178 John W. Du Bois
Berman, R.A., Ragnarsdóttir, H. and Strömqvist, S. 2002. “Discourse stance.” Written Language
and Literacy 5: 255–290.
Berman, R.A. 2005. “Introduction: Developing discourse stance in different text types and lan-
guages.” Journal of Pragmatics 37: 105–124.
Besnier, N. 1993. “Reported speech and affect on Nukulaelae atoll.” In Responsibility and Evi-
dence in Oral Discourse, J.H. Hill and J.T. Irvine (eds.), 161–181. Cambridge: Cambridge
University Press.
Blanche-Benveniste, C., Bilger, M., Rouget, C. and van den Eynde, K. 1991. Le français parlé:
Études grammaticales. Paris: Éditions du Centre National de la Recherche Scientifique.
Bruner, J. 1986. Actual Minds, Possible Worlds. Cambridge, MA: Harvard University Press.
Chafe, W.L. 1993. “Prosodic and functional units of language.” In Talking Data: Transcription
and Coding in Discourse Research, J.A. Edwards and M.D. Lampert (eds.), 33–43. Hillsdale,
NJ: Lawrence Erlbaum.
Chafe, W.L. 1994. Discourse, Consciousness, and Time: The Flow and Displacement of Conscious
Experience in Speaking and Writing. Chicago: University of Chicago Press.
Clancy, P.M. 1999. “The socialization of affect in Japanese mother-child conversation.” Journal
of Pragmatics 31: 1397–1421.
Clift, R. 2006. “Indexing stance: Reported speech as an interactional evidential.” Journal of So-
ciolinguistics 10: 569–595.
Culioli, A. 1990. Pour une linguistique de l’énonciation. Paris: Ophrys.
Davies, B. and Harré, R. 1990. “Positioning: Conversation and the production of selves.” Journal
for the Theory of Social Behavior 20: 43–63.
Du Bois, J.W. 1991. “Transcription design principles for spoken discourse research.” Pragmatics
1: 71–106.
Du Bois, J.W. 2001. Towards a Dialogic Syntax. Ms., Department of Linguistics, University of
California, Santa Barbara.
Du Bois, J.W. 2002a. “Stance and consequence.” Paper presented at Annual Meetings of the
American Anthropological Association, New Orleans, LA, November 20–24.
Du Bois, J.W. 2002b. “Dialogic rhetoric: Stance and intersubjectivity in the syntax of engage-
ment.” Paper presented at Volkswagen Foundation conference on “Rhetoric Culture,” Jo-
hannes Gutenberg University, Mainz, July 4–7.
Du Bois, J.W. 2004. “Searching for intersubjectivity: ‘Too’ and ‘either’ in stance alignment.” Pa-
per presented at 25th Conference of the International Computer Archive of Modern and
Medieval English (ICAME), University of Verona.
wards and M.D. Lampert (eds.), 45–89. Hillsdale, NJ: Erlbaum.
Du Bois, J.W., Chafe, W.L., Meyer, C. and Thompson, S.A. 2000. Santa Barbara Corpus of Spo-
ken American English, CD, Part 1. Philadelphia: Linguistic Data Consortium.
Du Bois, J.W., Chafe, W.L., Meyer, C., Thompson, S.A. and Martey, N. 2003. Santa Barbara Cor-
pus of Spoken American English, CD, Part 2. Philadelphia: Linguistic Data Consortium.
Ducrot, O. 1972. Dire et ne pas dire. Paris: Hermann.
Ducrot, O. 1984. Le dire et le dit. Paris: Minuit.
Geertz, C. 1973. The Interpretation of Cultures. New York: Basic Books.
Goffman, E. 1981. “Footing.” In Forms of Talk. Philadelphia: University of Pennsylvania Press.

Goodwin, C. and Goodwin, M.H. 1992. “Assessments and the construction of context.” In Re-
thinking Context: Language as an Interactive Phenomenon, A. Duranti and C. Goodwin
(eds.), 147–189. Cambridge: Cambridge University Press.
Goodwin, M.H. 2006. The hidden life of girls: Games of stance, status, and exclusion. Oxford:
Blackwell.
Gumperz, J.J. 1992. “Contextualization and understanding.” In Rethinking Context: Language
as an Interactive Phenomenon, A. Duranti and C. Goodwin (eds.), 229–252. Cambridge:
Haddington, P. 2004. “Stance taking in news interviews.” SKY Journal of Linguistics 17: 101–
142.
Haddington, P. 2005. The Intersubjectivity of Stance Taking in Talk-in-Interaction. Ph.D. disser-
tation. English Department, University of Oulu: Oulu University Press.
Haddington, P. 2006. “The organization of gaze and assessments as resources for stance taking.”
Text and Talk 26: 281–328.
Hanks, W.F. 1990. Referential Practice: Language and Lived Space Among the Maya. Chicago:
University of Chicago Press.
Harris, Z.S. 1952. “Discourse analysis.” Language 28: 1–30.
Haviland, J.B. 1991. “‘Sure, sure’: Evidence and affect.” Text 9: 27–68.
Haviland, J.B. 1996. “Projections, transpositions, and relativity.” In Rethinking Linguistic Rela-
tivity, J.J. Gumperz and S.C. Levinson (eds.), 271–323. Cambridge: Cambridge University
Press.
Heritage, J. 2002. “Oh-prefaced responses to assessments: A method of modifying agree-
ment/disagreement.” In The Language of Turn and Sequence, C.E. Ford, B.A. Fox and S.A.
Thompson (eds.), 196–224. Oxford: Oxford University Press.
Heritage, J. and Raymond, G. 2005. “The terms of agreement: Indexing epistemic authority and
subordination in talk-in-interaction.” Social Psychology Quarterly 68: 15–38.
Hill, J.H. and Irvine, J.T. (eds.). 1993. Evidence and Responsibility in Oral Discourse. Cambridge:
Hobson, R.P. 1993. Autism and the development of mind. Hillsdale, NJ: Erlbaum.
Hunston, S. and Sinclair, J.M. 2000. “A local grammar of evaluation.” In Evaluation in Text:
Hunston, S. and Thompson, G. 2000. Evaluation in Text: Authorial Stance and the Construction
of Discourse. Oxford: Oxford University Press.
Itkonen, E. 2005. Analogy as Structure and Process. Amsterdam: John Benjamins.
Jakobson, R. [1957]1990. “Shifters and verbal categories.” In On Language: Roman Jakobson,
L.R. Waugh and M. Monville-Burston (eds.), 386–392. Cambridge, MA: Harvard Univer-
sity Press.
Jakobson, R. 1966. “Grammatical parallelism and its Russian facet.” Language 42: 398–429.
Jakobson, R. 1981. “Poetry of grammar and grammar of poetry.” In Selected Writings, Vol. 3:
Poetry of Grammar and Grammar of Poetry, 87–97. The Hague: Mouton de Gruyter.
Jefferson, G. 2004. “Glossary of transcript symbols with an introduction.” In Conversation
Analysis: Studies from the First Generation, G.H. Lerner (ed.), 13–31. Amsterdam: John
Benjamins.
Johnstone, B. (ed.). 1994. Repetition in Discourse: Interdisciplinary Perspectives. Norwood, NJ:
Ablex.
180 John W. Du Bois
Kärkkäinen, E. 2003a. “‘Is she vicious or dense?’: Dialogic practices of stance taking in conver-
sation.” In Santa Barbara Papers in Linguistics 12: Recent Studies in Empirical Approaches
to Language, T. Nakayama, T. Ono and H. Tao (eds.), 47–65. Santa Barbara: University of
Kärkkäinen, E. 2003b. Epistemic Stance in English Conversation: A Description of its Interac-
tional Functions, with a Focus on ‘I think’. Amsterdam: John Benjamins.
Kärkkäinen, E. 2006. “Stance taking in conversation: From subjectivity to intersubjectivity.”
Text & Talk 26: 699–731.
Kidwell, M. and Zimmerman, D.H. 2006. “‘Observability’ in the interactions of very young
children.” Communication Monographs 73: 1–28.
Kockelman, P. 2004. “Stance and subjectivity.” Journal of Linguistic Anthropology 14: 127–150.
Labov, W. and Waletzky, J. 1967. “Narrative analysis: Oral versions of personal experience.” In
Essays on the Verbal and Visual Arts, J. Helm (ed.), 12–44. Seattle: University of Washing-
ton Press.
Langacker, R.W. 1987. Foundations of Cognitive Grammar, Vol. 1: Theoretical Prerequisites.
Stanford, CA: Stanford University Press.
Langacker, R.W. 1991. Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin:
Mouton de Gruyter.
Lemke, J.L. 1998. “Resources for attitudinal meaning: Evaluative orientations in text semantics.”
Functions of Language 5: 33–56.
Linde, C. 1997. “Evaluation as linguistic structure and social practice.” In The Construction of
Professional Discourse, B.-L. Gunnarsson, P. Linell and B. Nordberg (eds.), 151–172. New
York: Addison Wesley Longman.
Linell, P. 1998. Approaching Dialogue: Talk, Interaction, and Contexts in Dialogical Perspectives.
Lucy, J.A. (ed.). 1993. Reflexive Language: Reported Speech and Metapragmatics. Cambridge:
Macken-Horarik, M. and Martin, J.R. 2003. Negotiating heteroglossia: Social perspectives on
evaluation (Special Issue). Text 23.
Martin, J.R. 2000. “Beyond exchange: APPRAISAL systems in English.” In Evaluation in Text:
Maynard, S.K. 1993. Discourse Modality: Subjectivity, Emotion, and Voice in The Japanese Lan-
guage. Amsterdam: John Benjamins.
Mead, G.H. 1934. Mind, Self, and Society. Chicago: University of Chicago Press.
Moore, C., and Dunham, P.J. (eds.). 1995. Joint attention: Its origins and role in development.
Hillsdale, NJ: Erlbaum.
Ochs, E. 1988. Culture and Language Development: Language Acquisition and Language Social-
ization in a Samoan Village. Cambridge: Cambridge University Press.
ity, J.J. Gumperz and S.C. Levinson (eds.), 407–437. Cambridge: Cambridge University
Press.
Peirce, C.S. [1885]1933. “Three kinds of signs.” In Collected Papers of Charles Sanders Peirce,
Vol. 3, C. Hartshorne and P. Weiss (eds.), Cambridge, MA: Harvard University Press.
Pomerantz, A. 1984. “Agreeing and disagreeing with assessments: Some features found in pre-
ferred/dispreferred turn shapes.” In Structures of Social Action, J.M. Atkinson and J. Heri-
tage (eds.), 57–101. Cambridge: Cambridge University Press.
Recanati, F. 1989. “The pragmatics of what is said.” Mind and Language 4: 295–328.
Sacks, H. 1992. Lectures on Conversation. Oxford: Blackwell.
Schegloff, E.A. 1996. “Confirming allusions: Toward an empirical account of action.” American
Journal of Sociology 102: 161–216.
Schutz, A. 1962. Collected Papers I. The Problem of Social Reality. The Hague: Nijhoff.
Searle, J.R. 1983. Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge
University Press.
Shoaps, R.A. 2002. “‘Pray earnestly’: The textual construction of personal involvement in Pen-
tecostal prayer and song.” Journal of Linguistic Anthropology 12: 34–71.
Shoaps, R.A. 2004. Morality in Grammar and Discourse: Stance-taking and the Negotiation of
Moral Personhood in Sakapultek (Mayan) Wedding Counsels. Ph.D. dissertation. Linguis-
tics Department, University of California, Santa Barbara.
Silverstein, M. 1976. “Shifters, verbal categories, and cultural description.” In Meaning in An-
thropology, K.H. Basso and H. Selby (eds.), 11–55. Albuquerque: School of American Re-
search.
Silverstein, M. 1984. “On the pragmatic poetry of prose: Parallelism, repetition, and cohesive
structure in the time course of dyadic conversation.” In Meaning, Form, and Use in Con-
text: Linguistic Applications, D. Schiffrin (ed.), Washington, DC: Georgetown University
Press.
Silverstein, M. 2001. “The limits of awareness.” In Linguistic Anthropology: A Reader, A. Duranti
(ed.), 382–401. Oxford: Blackwell.
Sperber, D. and Wilson, D. 1995. Relevance: Communication and Cognition. Oxford: Black-
well.
Takanashi, H. 2004. The Interactional Co-Construction of Play in Japanese Conversation. Ph.D.
dissertation. Linguistics Department, University of California, Santa Barbara.
Tannen, D. 1987. “Repetition in conversation: Towards a poetics of talk.” Language 63: 574–
605.
Thompson, S.A. and Mulac, A. 1991. “A quantitative perspective on the grammaticization of
epistemic parentheticals in English.” In Approaches to Grammaticalization, Vol. 2: Focus on
Types of Grammatical Markers, E.C. Traugott and B. Heine (eds.), 313–329. Amsterdam:
John Benjamins.
Thompson, S.A. and Hopper, P.J. 2001. “Transitivity, clause structure, and argument structure:
Evidence from conversation.” In Frequency and the Emergence of Linguistic Structure, J.L.
Bybee and P.J. Hopper (eds.), 27–60. Amsterdam: John Benjamins.
Tomasello, M. 1999. The Cultural Origins of Human Cognition. Cambridge, MA: Harvard Uni-
versity Press.
Tomasello, M., Carpenter, M., Call, J., Behne, T., and Moll, H. 2005. “Understanding and sharing
intentions: The origins of cultural cognition.” Behavioral and Brain Sciences 28: 675–735.
Traugott, E.C. 1989. “On the rise of epistemic meanings in English: An example of subjectifica-
tion in semantic change.” Language 65: 31–55.
Voloshinov, V.N. [1929]1973. Marxism and the Philosophy of Language. Cambridge, MA: Har-
vard University Press.
182 John W. Du Bois
Vygotsky, L.S. 1986. Thought and Language. Cambridge, MA: MIT Press.
Wittgenstein, L. 1953. Philosophical Investigations. New York: MacMillan.
Appendix: Transcription conventions
The transcription symbols and conventions used in this paper are largely as in Du Bois et al.
(1992), although there have been a number of significant updates in my more recent transcrip-
tion practice. The most relevant symbols are given below. (For further details see http://www.
linguistics.ucsb.edu/projects/transcription/representing).
Meaning Symbol Comments

intonation unit line one new line for each intonation unit
speaker/turn attribution JILL; semicolon follows name in CAPS
simultaneous speech [ ] brackets show overlap start and end
pause, timed (1.2) pause duration in seconds
hold (micro-pause) .. less than 150 milliseconds
lag (prosodic lengthening) : colon marks slowing of local tempo
breath (H) audible inhalation
laugh @ one symbol for each pulse of laughter
unintelligible ### one symbol per unintelligible syllable
uncertain hearing #you’re #kidding transcribed words are uncertain
pseudograph ~Jill name change to preserve anonymity
The role of I guess
in conversational stancetaking
Elise Kärkkäinen
University of Oulu
1. Introduction1
My overall starting-point in this paper is an epistemic stance marker that has

been strangely neglected in linguistic research so far, namely I guess. Here, I draw
from evidence provided by recent linguistic work that has established epistemic/
evidential stance marking to be a highly regular and routinized phenomenon in
terms of the linguistic forms used: I guess belongs to this group of frequent mark-
ers. In Kärkkäinen (2003), I established that American English speakers use a lim-
ited set of high-frequency markers in everyday speech for their expression of epis-
temic/evidential stance.2 It is therefore conceivable that many of these frequent
stanced items have specialized into some routine function(s) in the interactional
organization of conversation. This is the case with another frequent expression, I
think, which in everyday American English has been shown in first-position turns
to routinely frame an upcoming stanced turn or longer opinion sequence, or in
second-position turns to project that a different speaker perspective will follow,
while it may also do less routine-like recipient design and face-work at points of
trouble in interaction (Kärkkäinen 2003; see also Rauniomaa this volume for es-
sentially similar findings concerning the Finnish minun mielestä ‘in my opinion’
and minusta ‘I think’).
In what follows, I will view I guess, an epistemic/evidential fragment that is
very frequent in American English (but not in British English), in an overall theo-
retical framework proposed by Du Bois (2001, 2003, this volume) of stance as
a social act that is achieved dialogically and publicly, through our engagement
with our co-participants. This view is not unlike the view held in discursive social
psychology, of evaluations as not carried around ready-made by participants, but
worked up in a way that is suitable for what is being done in interaction (see e.g.,
184 Elise Kärkkäinen
Potter 1998). In this paper, I indeed view stance essentially as emerging from dia-
logic interaction between interlocutors (Holquist 1990; Voloshinov [1930]1973;
see also Kärkkäinen 2006), and I am primarily interested in the practices and pro-
cesses of (the activity of) stancetaking between co-participants. I draw on conver-
sation analysis as a methodological tool to examine in detail the conversational
actions that I guess frames, and how those actions are designed linguistically and
prosodically. In this paper, I argue that I guess, a subjective marker par excellence
(but see Section 3.2), functions as an intersubjective stance frame that organizes
the stancetaking activity between conversational co-participants in a surprisingly
consistent fashion. By frame, I here mean a fragment of speech that provides a
perspective or stance toward the action that is produced in the associated utter-
ance (see Section 3.2 for a more detailed definition of frame).
It is significant that when we look beyond the initial frame, I guess, the rest of
the utterance typically also contains explicit stance-indicating or stance-evoking
material, such as evidential markers, markers of epistemic modality and evaluative
lexis. The speaker thereby frequently produces an action such as an assessment,
an opinion, or a (strong) claim, that inherently involves taking a stance or a posi-
tion. Indeed, Ochs and Schieffelin (1989: 22) have alerted us to the possibility that
affect (or stance) may permeate the entire linguistic system. The authors argue
that the linguistic resources for expressing affective and epistemic stance include,
not only the lexicon, but the following: grammatical and syntactic structures such
as choice of pronouns, determiners, verb voice, tense/aspect, sentential adverbs,
hedges, cleft constructions, diminutives, augmentatives, quantifiers, and word or-
der; phonological features such as intonation, voice quality, sound repetition, and
sound symbolism; and discourse structures such as code-switching as instantiated
by taboo words, dialect, couplets, and repetition of own/other’s utterances (Ochs
and Schieffelin 1989: 12-14; Ochs 1992: 412). Students of language have further
acknowledged the necessity and difficulty of identifying evoked or implied evalu-
ation, or cases where stance is not explicitly inscribed but seems nevertheless to
be evoked (Martin 2003; see also Clift 2006 for interactional evidentials which
are dependent on sequential position to index stance, as opposed to stand-alone
evidentials which inherently index a stance).
What, then, is the contribution of I guess in the conversational actions and
sequences that it appears in, if those actions need not be indexed or framed by I
guess to already count as stanced actions? The conversational data at hand suggest
that I guess brings in a special meaning of ‘making an inferential discovery,’ ‘com-
ing to a realization,’ or ‘drawing a conclusion,’ i.e., this marker displays and makes
public a (reasoning) process of the speaker at a particular point in interaction. But
if we view the action containing/framed by I guess as constituting one relevant
action within a longer sequence of stancetaking actions between co-participants, it
becomes pertinent to view I guess as not only indexing inference or conclusion,

but in effect organizing the stancetaking activity between the co-participants. As
we will see below, I guess can be viewed as a discursive practice used for perform-
ing a practical task in discourse (cf. Potter and Hepburn 2003): in sequence-ini-
tiating actions for displaying the discovery of a speaker’s stance, in responsive
actions for displaying and projecting that the speaker will give up on a stance
and/or adopt a new one, and in extended turns to insert a stance relevant for the
understanding and appreciation of a telling.
If some of the most frequent “traditional” stanced items in a language, such as
epistemic I think and I guess, turn out to essentially only act as stance frames, i.e.,
to display and project an upcoming stanced action and organize the stancetaking
activity between participants, rather than themselves indexing an interactionally
strong and salient stance on many occasions, it is essential that we acquire some
breadth in our descriptions of the manifestations of stance in naturally-occurring
discourse and language use. We clearly need to also look beyond these frequent
and routinized stance markers (such as markers of epistemic modality and evi-
dentiality, evaluative lexis, and affect markers) that linguists have largely occupied
themselves with. Indeed, my aim in this paper is to go deeper into the more in-
teractional resources of stancetaking. I wish to shed some light on the practices
and processes of stancetaking beyond individual speakers’ contributions, across
speakers and speaking turns (cf. repetition of own/other’s utterances mentioned
above).
I will therefore expand the scope of examination from the actions and turns
containing I guess, to the turns-at-talk coming before and after the current turn,
and ultimately to the larger activity that the participants may be seen to orient to
in a given situation. In other words, attention will be paid to the multiple ways
that stances are taken in turns-at-talk containing this initial stance frame, but
also in prior and subsequent turns. A frequent resource used thereby for taking
stances is the syntactic, semantic, and prosodic resonance between contributions
by different speakers, as proposed by Du Bois (2001; see also Anward 2000).
2. Data and transcription
My database consists of 54 occurrences of I guess gleaned from 17 different con-

versations of about 15–25 minutes each, from Parts I and II of the Santa Barbara
Corpus of Spoken American English (Du Bois et al. 2000; Du Bois et al. 2003).
Thus, some 50 sequential environments of I guess were analyzed in detail. One ex-
ample is taken from a corpus of British and American television news interviews
(Haddington 2003).
The data are transcribed according to the discourse transcription conventions

presented by Du Bois et al. (1993; see the Appendix for a list of symbols used), with
the intonation unit (IU) as the basic unit of transcription. The way we as analysts
transcribe our data, of course, already affects our perception of how speech and
interactional actions are produced and emerge through time. It is advantageous
to transcribe data into IUs, because the very visual representation, with each IU
lined up on a separate line, immediately alerts us to the emergent, incremental,
and directional nature of turns-at-talk and the turn-construction units (TCUs)
that turns are made of. Thus, the construction of a turn is frequently incremental,
or produced piece-by-piece rather than as one long chunk (Schegloff 1996: 55).
All TCUs are also directional, available to participants as they unfold, and not
only once they have been produced and completed (Lerner 1996: 307).
However, I have modified the Du Bois et al. (1993) convention in one respect.
The marking of pauses is generally done in accordance with the conversation ana-
lytical methodology: pauses appear on a separate line rather than at the beginning
of a certain speaker’s IU, to alert the analyst to the fact that it is frequently possible
to assign the pause to some other speaker instead.
3. I guess from a linguistic point of view
Recent research in functional and interactional linguistics has acknowledged the

need to view certain recurrent lexico-syntactic structures and discourse patterns
from a new vantage point, in order to better account for their use in actual ev-
eryday and institutional interactions (see e.g., Ford et al. 2002; Keisanen this vol-
ume). The very frequency of I guess makes it a worthy object of such study, and I
will here give a brief description of this collocation as a linguistic phenomenon,
even though most studies to date only deal with it in passing.
3.1 Frequency of I guess in spoken English
Several studies have pointed out that I guess is a very frequent phrase in everyday
spoken American English, even though not quite as frequent as I think or I don’t
know.
In Kärkkäinen (2003), personalized stance markers that make reference to
the speaker, such as I think, I guess, I know, I feel (like), and I found, were by far the
most frequent epistemic markers. This finding gains strong support from some
other recent studies based on spoken language corpora, namely those of Biber et
al. (1999), Thompson (2002), and Scheibman (2001, 2002).
Biber et al. (1999: 667–669) found that in a large database of some 6.4 million
words of conversational American and British English, the most common verbs
controlling that-clauses are the following:
think 2,000 per million words3
say 1,250 per million words
know 750 per million words
guess 500 per million words (in American English only)
In fact, the statistics provided in the Biber et al. (1999) grammar show a dramatic
drop in frequency after the epistemic verb guess: the next verbs, see, find, believe,
feel, suggest, and show only occur less than a hundred times per million words of
talk. This can be taken as evidence for my claim of the relatively small size of the
set of frequent stance markers.
Thompson (2002: 138) also observes that the most common complement-tak-
ing predicates (CTPs) in her data are the following, the great majority of which
occur with first-person subjects as fixed formulas:
think/thought 139
know/knew 51
see/saw 17
guess 17
remember 15
In Scheibman’s (2001, 2002) data, I guess was also highly frequent, along with the
two other epistemic phrases I think and I don’t know.
For comparison, I conducted a search of the three most common epistemic
phrases in the conversational data of the International Corpus of English–Great
Britain (ICE-GB) database, or 100 face-to-face and telephone conversations of
5–20 minutes in length (totaling 205,608 words).
I think 729 tokens or 0.35 % of 205,608 words
I don’t know 309 tokens or 0.15 %
I guess4 20 tokens or 0.01 %
We can see here the same overall order of frequency between the three phrases as
in the above American English data sets. As this search yielded 20 occurrences of
I guess, the number of I guess per million words in these data would very roughly
amount to 100, which is considerably less than the 500 in the Biber et al. (1999)
database. But we may conclude that even though this item is much more frequent
in American English, it is not entirely unheard of in everyday British English.
Here, however, I confine myself to American English usage.5
3.2 The fragment I guess as a stance frame
As becomes obvious from many recent linguistic studies, I guess is commonly

regarded as a conventionalized, sedimented pragmatic expression, i.e., a stance
marker, which frequently precedes complement clauses in English (Scheibman
2002: 32). Complement clauses have commonly been regarded as subordinate to
the main clause, i.e., to an epistemic phrase like I think, I (don’t) know, and I guess
(see Thompson 2002 for a summary of studies that hold this view).6 Thompson
and Mulac (1991a, 1991b) claim I think and I guess to have grammaticized into
epistemic phrases that act in the same way as adverbs, i.e., as adverbials, and be-
have much like epistemic morphemes in other languages. They note, however,
that it is not clear what the grammatical status of an epistemic phrase is.
In a later study, Thompson (2002) goes on to argue that we might term the
frequent CTPs – I think, I thought, I guess, I remember, and I know/knew – epis-
temic/evidential/evaluative fragments, rather than main clauses obtaining a com-
plement clause. Thompson further claims that the most frequent of these phrases
become relatively fixed epistemic formulas (e.g., I think/I don’t think/I thought/I
didn’t think), notably with first-person subjects, while the less frequent ones show
more diversity of form (e.g., make sure, tell, be interesting). Scheibman (2001: 70–
71, 76) also found that there are highly frequent formulaic collocations of first-
person-singular subjects and especially verbs of cognition (I guess, I don’t know,
I think) in her data, and as was mentioned above in Section 3.1, this finding is
corroborated in Biber et al. (1999: 667–669).7
It is worth pointing out, however, that the actual degree of subjectivity and
referentiality of frequent epistemic fragments like the ones above can be quite
low. Scheibman (2001) claims that there is a weakness or generality of referential-
ity in the subject of expressions like you know, I mean, and I think: the speaker
is not specifically referring to himself in these usages, but the entire collocation
takes on pragmatic function with a concomitant reduction in specificity of the
pronominal referent. Traugott (1995: 38) further claims of the English I think that
it is becoming more subjective both in function (toward a fixed phrase indicat-
ing speaker’s epistemic attitude) and also in the overwhelming selection of the
first-person-subject form, which may eventually become eroded and leave only
a discourse particle (presumably think). We may also ask how much referential
meaning is left in the predicate and ultimately in the whole collocation. Schif-
frin (1987) argues that some cases of discourse markers like y’know are less ref-
erentially meaningful than others (1987: 319), while Stenström (as cited in Helt
1997) observes of I think, I mean, you know, and you see that they fill various slots
along a semantic continuum that represents a strong relationship to their literal
meanings at one end, and a minimal relationship to literal meaning at the other.
Lenk further points out that the (short) lexical items that may figure as discourse
markers, i.e., that are used to “signal the sequential and ideational relationship
of the two utterances between which they occur, or to other segments within the
discourse” (1998: 52), often also have a (separate) propositional meaning (cf. to
guess the answer). The collocation I guess, however, mostly appears with a prag-
matic meaning in my data, even though it still carries remnants of its referen-
tial or propositional meaning in many contexts of use.8 Indeed, in Kärkkäinen
(2003: 175–179), I established the discourse-marker status of I think in everyday
spoken English, and in the present paper we will note clear evidence for a similar
status of I guess: it may operate at both a local and a global level in discourse, it is
syntactically detachable from sentences, it commonly appears in initial position
of an utterance, it may have a range of prosodic contours, and either has no mean-
ing or only a vague meaning (see Schiffrin 1987 for criteria for discourse markers,
to be discussed in more depth in Section 5).
Let us then look at the following example, presented in Thompson
(2002: 132):
(1) (at a birthday party, after Kevin was discovered to have lettuce on his tooth,
everyone has jokingly commented on it, and Kendra has asked for a toothpick)
WENDY: ... everybody’s getting uh,
tooth obsessed.
KEN: I guess we a=re.
Thompson draws our attention to the fact that Ken’s aligning agreement to Wen-
dy’s summary of the previous turns is expressed in the complement clause we
are, and that it is not the complement-taking phrase, here I guess, that consti-
tutes the actual action of the turn. She ends up suggesting (2002: 142, 146) that
what has been termed the main clause in much linguistic research simply serves
as a stance frame for the clause that it occurs with: such epistemic/evidential/
evaluative frames or fragments provide a certain type of perspective or stance
toward the actions, i.e., the assessments, claims, counterclaims, and proposals,
being done in the associated utterance. In effect, Thompson (2002: 141) proposes
a view of grammar as reusable fragments, to be used as turns or parts of turns,
i.e., as practices of turn construction. This is very much the view that I also adopt
here, and in what follows I will examine what exactly the role and function of the
frame I guess is in turn construction and more specifically in stancetaking activity
between participants. I guess will be examined in connection with different kinds
of conversational actions done in different sequential environments: in initiating
and responsive actions, and in extended or multi-unit turns.
3.3 A note on the semantic meaning of I guess
Few studies actually deal with the semantics of this stance marker. Chafe (1986),
however, notes that I guess is a marker of “belief.” He further claims (1986: 266)
that belief is a mode of “knowing” in which concern for evidence is downgraded,
and speakers believe things because other people do or simply because they want
to believe them. Thompson and Mulac (1991a: 325) observe in passing that, as
epistemic phrases, I think and I guess have still retained vestiges of their earli-
er lexical meaning in the grammaticized form: the two still have a difference in
meaning, so that I think is a stronger assertion of belief than I guess. This, the au-
thors argue, is “traceable to the difference between think and guess as verbs: guess
implies an assertion based on little or no evidence, and hence less commitment to
a proposition than think does” (1991a: 325).
But on a closer analysis, I guess in my data frequently appears in contexts
where the speakers do show an orientation toward evidence, which is presented
in, or inferable from, the immediately prior turns, or is even present in the actual
physical or wider social environment where talk takes place. Here, it differs from
I think, which in semantic orientation actually expresses points on a continuum
from speaker uncertainty to relative certainty (even though it appears in very
similar sequential environments, see Kärkkäinen 2003). It turns out that I guess
is actually used as an evidential rather than an epistemic marker, as will become
clearer from many of the examples that I will now turn to.9
4. Practices of stancetaking: Beyond the stance frame I guess
4.1 I guess in sequence-initiating actions: Discovering a stance
I will here examine cases in which I guess frames sequence-initiating actions, such
as (first) assessments, opinions, assertions, and questions, which are followed by
second actions or second-pair parts, such as second assessments, opinions, as-
sertions, and answers. The following examples show that I guess marks actions in
which some kind of change in the speaker’s current state of knowledge or aware-
ness or orientation has (just) taken place (cf. Heritage 1984 for the change-of-state
token oh). The stance displayed frequently arises from an inference made by the
speaker, and is based on evidence gleaned from a just prior turn or turns, from the
actual environment where talk takes place, or from the wider social context. Such
a stance, as any act of stancetaking, necessarily invokes an evaluation at some lev-
el, whether this is actually asserted or otherwise inferable (Du Bois this volume;
see also Martin 2003 for evoked evaluation). ‘Evaluation’ in turn means charac-
terizing the stance object as having some specific quality or value (Du Bois this
volume; see also Goodwin and Goodwin 2002: 154 for assessments as evaluations
of persons and events being discussed within talk). And, finally, actions involv-
ing I guess propose something as a possible stance object that the co-participants
can subsequently position themselves towards, i.e. such actions frequently initiate
stancetaking activity between the conversational co-participants (see also Good-
win and Goodwin 2002 for assessment activity as something that the participants
orient to, and participate in, together).
Displaying a sudden change in the speaker’s state of knowledge or aware-
ness is frequently witnessed in first-position turn formats of the following type
in which the speaker displays on-line that his or her immediately preceding ut-
terance was somehow wrong or mistaken and no longer relevant, and in effect,
cancels out the need for the action performed in it, typically a question:
(2) (Wonderful Abstract Notions SBCSAE 0017 <00:19:44>)

1 MICHAEL: ... So they’re `color-cycling right ^now.
2 ... Does it ^zoom `later?
3 [I `guess it ^must].
4 JIM: [(H) Yeah,
5 Ì’ve got] ^zooms on here `too.
While producing the question in line 2, Michael seems to reason and infer, on the
basis of how videos are usually designed, that sooner or later the picture will zoom
in on something. The semantic meaning of must ‘induction’ (Chafe 1986) of course
strengthens this interpretation, yet the second part of such a turn format may also
contain other types of predicates (or even no predicate at all: Is @Yoyo @Ma Chi-
nese? .. I guess with a name like Yoyo=; SBCSAE 0019). Notice how interactive this
turn design actually is: the speaker engages in interaction with him- or herself, and
simultaneously makes public his or her inferential process and discovery of some
state of affairs.10 Such turns are also regularly responded to by the recipient, who
may legitimately speak at the end of the first turn-construction unit, the actual
question, even though the current speaker also continues to speak.
A further example of a change in the speaker’s awareness, or the speaker mak-
ing a discovery based on inference, is provided by the following example (3). The
topic of prior discourse has been Harold’s nephew Thomas, a three-year-old who is
now learning to tap dance. The boy’s parents had been inspired by a famous young
tap dancer, who is “awesome” and incredibly fast. Prior to this extract the partici-
pants have been discussing and assessing him from their respective angles.
(3) (Lambada SBCSAE 0002 <00:00:54>)

1 HAROLD: I’m sure `Thomas is all ôver it.
2 (3.7)
3 JAMIE: Prob[ably XX][2XXX2] --
4 HAROLD: [<F><HI> I mean `he][2has a bro-2</HI><F/>] --
5 MILES: [2XXXX could have2] ^see=n `him.
6 HAROLD: I `guess that `means his broken `leg is [3@doing @ôkay3].
7 PETE: [3I was ^wonder3]ing
about `that,
8 I was ìmagining [4he had `broke an ârm4] or something.
9 JAMIE: [4<HI> Oh yeah= </HI>4].
10 PETE: But it was his ^leg?
Harold brings the prior topic to a close on line 1 by offering a concluding evalu-
ative claim or assessment concerning the little nephew’s enthusiastic attitude to-
ward tap dancing. There is clearly a topic closure here (cf. Goodwin and Goodwin
1992: 169), which is marked prosodically by Harold’s lowered volume and a long
pause after this turn. Then several participants self-select to speak almost simul-
taneously, but Harold’s new beginning is clearly signaled by a rather loud and
high-pitched I mean `he has a bro- -- on line 4; he refers to the nephew, as will
become obvious a little later. This is overlapped by Jamie (line 3, largely inau-
dible) and Miles (line 5, very likely talking about the nephew as well, wishing that
he could have seen him tap dance). On line 6 Harold redesigns his turn (rather
than simply restarting and repeating what he had set out to say) and produces an
inference I `guess that `means his broken`leg is @doing @ôkay., which already
incorporates as a presupposition that the nephew has in fact broken a leg. The
broken leg had not been topicalized in the conversation so far, even though we
can now be fairly certain that Harold intended to announce it as news on line 4, I
mean he has a broken leg. On line 6, he produces the action of simultaneously (in
passing) announcing a piece of news and evaluating it (doing okay), a turn design
that highlights the unexpected nature of the state of affairs. Harold also produces
the last two words with some laughter, thereby inviting laughter from the other
participants (Jefferson 1979). Pete and Jamie decline this invitation and take up
the topical import of Harold’s talk instead. Their subsequent turns display that
Harold has indeed drawn their attention to something that they had not thought
of or fully realized during the discussion so far, even though they clearly had some
prior knowledge of the boy’s broken leg.
The role of I guess – or in this case I guess that means – here is to project and
frame the speaker’s recent inference or realization, which he is about to put on the
table. I guess signals that this particular stance is such that it arises from an almost
on-line reasoning process of the speaker. When producing such an inference, the
speaker digresses from his original trajectory: rather than simply self-repairing
and re-producing his original news announcement, he proceeds to evaluate the
newsworthy item, the broken leg. We can therefore say that stance and evalua-
tion take the center stage here, and thereby affect the original action trajectory
and sequential organization of talk. What is more, by introducing such a stance,
the speaker invites and involves the co-participants in stancetaking in the same
way as has been shown for one specific subtype of stanced actions, namely first
assessments, which make second assessments conditionally relevant (Pomerantz
1984). In example (3) the co-participants join in to further evaluate the nephew’s
recovery, and a long sequence follows during which the participants discuss and
assess how quickly the nephew had healed and how children’s bones in general
“grow back really fast”.
In the next example, the evidence is clearly seen in the environment, the In-
donesian masks that Harold and Jamie have hanging on their wall.
(4) (Lambada SBCSAE 0002 <00:07:05>)

1 MILES: ... <HI> `Those two top ^ma=sks </HI>,
2 .. `there.
3 (1.7)
4 HAROLD: ^Yep.
5 (0.8)
6 JAMIE: @@@
7 HAROLD: [We `need a ^verb].
8 MILES: [I ^guess the] [2ònes below are `women2],
9 PETE: [2@@@@@2]
10 JAMIE: [2@@@(H)2]
11 MILES: [3Is `that why they look so ^different3]?
12 JAMIE: [3@@ (H)3]
13 HAROLD: .. Ûh-oh.
14 MILES: I mean,
15 .. i%- --
16 (1.1)
17 Well you see their eyeballs,
18 [I guess].
19 PETE: [Right].
20 .. Yeah [2the other2] --
21 MILES: [2In addition2] to a mustache and a beard.
22 (1.0)
23 But the thing is,

24 [that second one looks like the guy] who was in [2one of the2]
[3Oba Oba3] skits.
Miles introduces the two top masks as the actual referential focus of talk on line
1, and only in lines 23–24 does he proceed to make the point that one of the two
top figures looks like a performer in a show that he has recently seen. But before
he initiates talk on that exact referent, he displays some trouble in continuing his
talk, as can be heard from the long pauses and the fact that Harold humorously
prompts him to continue (lines 4 and 7). In line 8 Miles then produces a noticing,
and engages in a brief comparison of the top masks and the ones hanging below
them; he appears to be struck by a difference between them and offers as a reason
that the two lower ones portray women rather than men. He also explicitly invites
his co-participants to take a stance by asking them to ratify the reason he has just
proposed on line 11: the question/request already presupposes as a given that
they indeed look different. The two utterances are clearly designed to be produced
together as one chunk: the noticing in line 8 (which ends in a continuing intona-
tion) acts as a necessary background stance for the more explicit stancetaking ac-
tion, the actual question, in line 11, which is said in a kind of latching prosody, in
an overall high pitch, continuing at the height where the previous intonation unit
left off. In what follows, both Jamie and Harold display that there is some trouble
involved in Miles’s proposed stance: Jamie laughs on line 12, and Harold displays,
by a mock alarm cry uh-oh, that there was something quite problematic in the
prior turn. Miles himself realizes soon enough that the women in the masks have
“a mustache and a beard”, that they do not portray women at all but male figures.
After this side sequence he resumes the original topic, the figure portrayed in one
of the top masks.
In sum, what we can say about the role of I guess here is that it again appears
in a context, in a side-sequence, where the speaker makes public an inference (a
noticing) that was based on some evidence provided by the immediate environ-
ment. At the same time, the insertion of such an inference takes place in a sequen-
tial position where the speaker rather abruptly changes the trajectory of his talk.
The inference is also part of a turn design that generally invites others to take a
stance on the matter, even biases them toward a certain stance.
We can see essentially a similar pattern in the following excerpt from insti-
tutional data, namely from an American television news interview recorded in
the aftermath of the 9/11 terrorist attacks. The same example is also discussed by
Haddington (this volume) with respect to the turn-design features of the inter-
viewer’s and the interviewee’s turn.
(5) CNN, Larry King Live, Sep 12, 2001: Bomb him out
IR: Larry King, IE: Brian Jenkins (001 / 5 / 3:49)
1 IR: (0.6)(TSK) ^Brian,
2 could `you=,
3 ...<PAR> well I `guess the `public,
4 would `look at this `simply </PAR>,
5 `Could you ^bomb him out.
6 IE: .. (TSK)(H)U=h,
7 I d- --
8 I don’t know that you could ^bo=mb him out,
The interviewer (IR) designs his turn on-line, as he re-directs it after having al-
ready initiated the question Brian, could you=. He subsequently inserts an utter-
ance which incorporates the stance of an ‘other,’ namely of the general public, who
are represented as favoring the rather extreme action of simply bombing the ter-
rorist leader out from wherever he is hiding (a rather widespread public opinion
established on some rapid opinion polls). Well here signals an orientation shift
within the speaker’s extended turn (Schiffrin 1987: 102–127), and displays to the
interviewee that the action that follows is not fully coherent with the main trajec-
tory of the question. It is further framed by I guess, which projects an upcoming
stance, in this case a third-party stance arrived at via an on-line realization of the
speaker (see Haddington this volume for a fuller description of such “position-
ing” activity). Iconically, this inserted stance is also produced in a parenthetical
prosody, namely in slightly lower pitch and faster tempo.
Goodwin (1981: 127) suggests that such modifications of the turn construc-
tion are regularly done by participants to “coordinate their production with the
actions of a recipient”. For example, a phrasal break and subsequent insertion of
a new unit (a word, a phrase, or a whole sentence) in the emergent turn is often
caused by the fact that the speaker has not yet secured the recipient’s gaze or
needs to otherwise negotiate a state of mutual focus with, and coordinate his or
her actions with those of, the recipient. Goodwin (1981: 127) gives the following
example that contains the insertion of I guess at precisely the point where the gaze
of the recipient (marked by X) arrives:
↓
GARY: He’s a policeman in Bellview and he :, I guess he’s,
[
X__________
In example (5) above, however, something else is going on. The IR engages in
looking at the IE only until the end of line 2. Simultaneously with the onset of
the parenthetical digression starting with well I guess, he withdraws his gaze from
the IE and looks down. He also turns his palm upward, possibly in acknowledge-
ment that he had already brought up the results of the opinion poll earlier in the
interview, or as an ‘offer’ or a ‘presentation’ of the discourse (or stance) object to
the recipient (Kendon 2004: 123). Only after the insertion is completed does he
turn his gaze toward the IE again, while the palm remains in its upward position.
The embodied production of the insertion, especially the downward gaze and
the (parenthetical) prosodic production, together contribute to its interpretation
as an interactionally engendered ‘remembering’ (Heritage 2005), a realization of
the IR inserted in the larger question sequence. On the other hand, it is possible
to argue that the withdrawal of gaze works towards ensuring that the speaker can
keep the turn during the production of the insert and subsequently continue with
the projected trajectory of the turn (cf. Lerner 1991).
The above insertion again causes talk to diverge from its original trajectory.
We may conclude that I guess acts as a frame in a turn that is designed to convey
a third-party stance inserted on-line, rather than the actual stance of the current
speaker himself. We can find hints as to how the IR wishes to present himself in
view of the matter at hand: he had started out by saying could you=, on line 2,
very likely resulting in could you bomb him out, which would have constituted a
valid question presented in the IR’s “own name.” By inserting the I guess-framed
digression, the IR switches from this to proposing another stance that he may or
may not himself share, and to now asking the question “on behalf of ” the general
public rather than himself. Finally, this turn design invites the IE to take a stance
on the matter and, while doing so, take into account the ostensibly simple solu-
tion favored by the general public (but not necessarily shared by the IR).
In sum, I guess in initiating actions like the ones above indexes a just discov-
ered stance, a sudden change in the speaker’s epistemological state of knowledge
or awareness or orientation. This stance arises from a reasoning process of the
speaker, which is often based on evidence or some other stimulus gleaned from
just prior turn(s) or even the ongoing one, from the actual environment where
talk takes place, or from the wider social context. I guess here comes quite close
to another marker of evidentiality and stance in British English data, surely, the
use of which is claimed by Downing (2001) to be triggered by a psychological
event of “coming into awareness” at the very moment of speaking. This evidential
form is said to index information that is new or unexpected to a speaker, while
the stimulus for this is derived most likely from the verbal message but also from
visual or sensory evidence, or via inference (Downing 2001: 277). The difference
between I guess and surely in the degree of change in awareness is then traceable
to the semantic meanings of guess and sure. A special flavor of I guess, not unlike
that of surely, is that the stance that “surfaces” in the associated utterance often has
a humorous and surprised quality, and sometimes comes close to sarcasm.
What is essential is that even though we see evidence of some private pro-
cesses of inference in the turns framed with I guess, these nevertheless derive from
and surface in the intersubjective to and fro of dialogue. The I guess-framed ut-
terance often constitutes a clear side sequence within the current speaker’s ongo-
ing sequence, initiated and closed by the speaker him- or herself (Lerner 1991;
Local 1992; Duvallon and Routarinne 2005; cf. also Jefferson 1972 for subsidiary
side sequences initiated by another speaker). I guess can further be seen to or-
ganize the stancetaking activity of the co-participants, as the actions containing
this frame often invite them to also take a stance in subsequent discourse, and
in doing so to take into account the stance already implicit or explicit in the de-
sign of the ongoing turn. We may also argue that at such points in interaction,
stances gain in importance to such an extent that they impinge on the sequential
organization of talk and in some way change or temporarily withhold its original
(action) trajectory.
4.2 I guess in responsive actions: Giving up and redefining a stance
Let us now turn to cases where I guess appears in second position (i.e., in respon-
sive actions to some other actions) in conversational sequences that often involve
some degree of disagreement and disaffiliation between the participants. Here, I
guess similarly acts to display and project – on the basis of just prior discourse –
that the current speaker wishes to modify, withdraw, and redefine his or her origi-
nal stance at this point, i.e., to align with another speaker and, frequently, to dis-
play affiliation and a convergent stance. Note that even though many researchers
use the terms ‘alignment’ and ‘affiliation’ almost interchangeably, in what follows,
by aligning action is simply meant one that is produced within the trajectory of
the sequence-so-far (e.g. as one in an adjacency pair sequence), while an affiliat-
ing action is used to refer to one that establishes and increases social solidarity
between the social actors involved.
In the next example, Ken is telling a story about his childhood, when he had
witnessed the feeding of a fish, an Oscar, in a pet store: a goldfish was put in
the tank and the Oscar started chasing it. Joanne, as one of the two recipients, is
taking an active role at points in the telling (Joanne and Ken are a couple), even
though she has no prior knowledge of the incident.
(6) (Deadly Diseases SBCSAE 0015 <00:24:14>)

1 KEN: and `then= uh,
2 (H) But the `goldfish got ^s=tuck.
3 .. <MRC> ^half-way înto his ^mou[=th </MRC>].
4 JOANNE: [It `went the] wrong ^way.
5 KEN: ^No,
6 it [was `going],
7 JOANNE: [^Fin `first]?
8 KEN: it was `going,
9 JOANNE: .. ^Mouth `first?
10 KEN: .. Well it was go --
11 went ^tail `first.
12 JOANNE: `That’s the [^problem.
13 KEN: [I `guess that’s the ^wrong way,
14 JOANNE: (H) `That’s the <HI> ^problem </HI>].
15 KEN: <A> Ì don’t `know </A>.
16 (H)] Ânyway,
17 JOANNE: [2(H)2]
18 KEN: [2(H)2] .. (TSK)`So,
19 (H) you got this Ôscar there,
20 `swimming around [in the ^tank,
There is a danger from line 4 onwards that Joanne may halt or derail the storytell-
ing (cf. Mandelbaum 1989: 118). She hastens to propose a reason why the goldfish
might have been stuck in the Oscar’s mouth, It `went the wrong ^way. Ken starts
to disagree on lines 5–6 (^No, it was `going,) and again in line 8 (it was `going,),
but he is interrupted by Joanne, who proposes further candidate alternatives for
how the fish was going (^Fin `first?, line 7 and ^Mouth` first?, line 8). For Joanne,
it is of some consequence which way the fish was going: this story may in actual
fact confirm her just prior story about a snake that was likewise fed with a gold-
fish, her point having been that the snake would always eat the goldfish by the
head. She therefore assumes that the problem in Ken’s story was that the Oscar
was eating the fish by the fin (which for her equals tail, as she has been talking
about the two interchangeably) rather than by the head, and actively pursues that
in the discourse that follows.
On lines 10–11, Ken displays recognition that the fish actually went in tail
first. His turn design conveys that he comes to this conclusion pretty much at the
moment of speaking: he self-repairs and restarts with a simple past tense in line
11, went ^tail `first., which is produced with a drop in pitch on tail that indicates
some resignation at the realization of this fact. For Joanne, this of course proves
her earlier point about the snake, and she rather enthusiastically displays this by
producing `That’s the ^problem, with the pitch going up and the pitch range wid-
ening during the second production of <HI> ^problem </HI> (partly also be-
cause she is overlapped by Ken).
Ken finally concedes on lines 13 and 15 that Joanne may be right, I `guess
that’s the ^wrong way, <A> Ì don’t `know </A>. This turn indicates that Ken is
giving up on what for him is possibly a relatively small detail in the bigger scheme
of things, in order to retain his role as the main storyteller and to continue with
his story. There is some evidence for this in how Ken designs his turn. He hastens
to agree already upon hearing Joanne’s that’s the (line 12); this particular syntactic
chunk or collocation strongly projects that Joanne is about to produce an evalua-
tive turn, whether this is of a formulaic type (That’s the problem/point) or simply
ends by repeating Joanne’s earlier evaluative item the wrong way (cf. line 4). Ken
indeed displays an understanding of Joanne’s turn-so-far as an evaluative one,
even though he misprojects what type of evaluative item will follow (the problem),
and produces the wrong way instead. In fact, the participants have slightly shifted
their stance object here: Joanne orients toward gaining wider evidence for her
larger point (that eating a fish by the tail can be a problem not just for snakes but
also for predatory fish), while Ken still orients to the more local ‘which way is the
wrong way.’ Ken finally tags on I don’t know in a very accelerated form, thereby
marking that this side sequence in the story has come to a completion (cf. Ford
and Thompson 1996: 169; Scheibman 2000). After an audible inbreath he contin-
ues the story with Ânyway, (cf. Jefferson 1972: 316–320 for resumption of the
ongoing sequence; cf. Lenk 1998: 71–78 on anyway closing general conversational
digressions and resuming the earlier conversational topic).
The role of I guess in this sequence is to display that, in light of what was
established in just prior discourse, the speaker is giving up his original stance
and adopting a new one. In this sense, its use here is not far removed from that
witnessed in initiating actions (see Section 4.1), because here as well, the stimulus
or evidence that occasions the adjustment of the speaker’s original stance derives
from the immediately preceding turns (which again display the speaker’s on-line
recognition of some relevant facts).
In the following, I will present the turns above in which stances are being ne-
gotiated and a (more) convergent stance is achieved, in the form of two diagraphs.
Such a diagrammatic form of presentation has been proposed by Du Bois (2001),
to make visible the frequent modification of stances that takes place between dis-
course participants: a stance is often a “product of an immediately prior act of
stancetaking toward a shared stance object” (Du Bois 2003). Diagraphs map the
relevant intonation units produced by different speakers onto each other (rough-
ly) according to syntactic structure.
In Diagraph 1, the participants first establish which way the fish was going
in the Oscar’s mouth. In doing so, the speakers resonate both with their own and
with each others’ talk at several levels of linguistic structure.
Diagraph 1 (from example 6):

4 JOANNE: It `went the wrong ^way.
5 KEN: ^No,
7 JOANNE: ^Fin `first?
9 JOANNE: ^Mouth `first?
10 KEN: Well it was go --
11 went ^tail `first.
Joanne completes the syntactic structure initiated by Ken (lines 6–7 and 8–9).
This format is broken by Ken himself on line 11, where he engages in self-repair:
the design of this utterance displays his initial recognition of the fact that the
fish may indeed have gone in the wrong way. The past tense went signals finality
compared to the past progressive was going and resonates with Joanne’s original
wording on line 4, also in the past tense. He also produces the final item, ^tail first,
in resonance with both the syntactic form and prosodic realization (in terms of
accent) of Joanne’s candidate items in lines 7 and 9. Their versions of reality now
match factually (cf. Mandelbaum 1993: 263 for negotiations about versions of “re-
ality”), but what the implications of this are for their initial disagreement need to
be made more explicit. The next diagraph presents how this is done.

4 JOANNE: It `went the wrong ^way.
12 JOANNE: `That ’s the ^problem.
13 KEN: I `guess that ’s the ^wrong way,
Joanne’s initial anticipatory judgment of the fish going in ‘the wrong way’ was
expressed on line 4. After the participants have established together that the fish
had in fact entered the Oscar’s mouth ‘tail first,’ Joanne produces an explicit ‘I-
told-you-so’ concluding evaluation on line 12. Ken produces an affiliating second
evaluation on line 13, even though, as we have seen, slightly shifting the stance
object here. Here, we can see that Ken’s actual turns-at-talk are largely built on
Joanne’s previous talk and the stance displayed therein. Ken resonates with the
syntactic structure and evaluative terms in Joanne’s prior turns: he picks up the
reference term that and the predicate ’s from Joanne’s immediately previous turn,
and also recycles her earlier evaluative element, the wrong way. He further signals
his convergent stance by framing the turn with I guess.
The following is another instance of storytelling. Kevin and Wendy (a couple)
are jointly telling about an incident that they had witnessed together recently in
the parking lot of a mall; however, their interpretations of what exactly the pro-
tagonist was doing or trying to do differ.
(7) (Appease the Monster SBCSAE 0013 <00:12:33>)

1 KEVIN: ... Some `guy came òut and he- he was,
2 @[=]
3 WENDY: [<HI> Oh </HI>].
4 KEVIN: he was trying to `sell us ^cologne [2f-2]–
5 KENDRA: [2(COUGH)2] [3(COUGH)3]
6 WENDY: [2No2],
7 [3he `wasn’t3]
trying to ^sell [4us `co4]lo[5gne5],
8 KEVIN: [4`Well it-4]–
9 [5No5]=,
10 I <HI>`guess </HI> he was `trying to like,
11 ^lure us to a .. `place where they ^would `sell,
12 like,
13 .. îmitation `cologne,
14 but he `said,
15 <VOX> it’s `not îmitation,
16 `be[cause] </VOX>,
Even though in the immediately preceding story preface sequence Wendy had
aligned herself as a story consociate, here it is apparent that for her there is now
some trouble involved (cf. Lerner 1992). Kevin produces an assertion that frames
the story and the incident as having been mainly about selling, or about trying to
sell, Some guy came out and he was trying to sell us cologne (lines 1 and 4). This
turn design contains an implied evaluation, an evoked stance, namely that try-
ing to sell cologne to them is something rather strange, funny, or doomed to fail.
We can find evidence for this in Kevin’s production of this turn. He produces a
laughing particle in the middle, restarts several times as if to further emphasize
his upcoming point, and can be heard to put on a voice quality of “slight despera-
tion.” He is thereby orienting the other participants toward the upcoming story
in a certain way.11
But Wendy, the story consociate, disagrees on such a framing and puts in a
correction in lines 6–7. Wendy explicitly disaffiliates with Kevin, No he wasn’t try-
ing to ^sell us cologne, with an emphasis on the contentious item, sell. It is not the
case that Wendy is disagreeing on the facts of the story (cf. Lerner 1992 on story
consociates correcting facts), it is rather that their fundamental perceptions differ
of what actually happened in the parking lot, and their versions of “reality” are
being negotiated here (Mandelbaum 1993: 263).
Upon hearing the emphasized word sell, Kevin immediately starts to redesign
his turn; `Well it- -- displays that he may be about to start with a further disagree-
ment, but he then retracts and produces an explicit No=, in effect agreeing with
Wendy’s previous disaffiliating turn. In what follows, he produces a new claim
that takes back some but not all of what he had said initially, I guess he was try-
ing to like, ^lure us to a .. place where they ^would sell, like, imitation cologne. He
offers a new and more precise understanding of the past events, that the guy was
going to sell something later, in other words Kevin does not altogether withdraw
the idea of selling.
Kevin’s use of I guess is highly interactive in that it arises out of a need to
retract and redesign a turn to display a new and modified stance that converges
more with the previous speaker’s stance. It has a concessive flavor and could be
generally glossed as ‘In light of what you just said there is enough evidence for
me to modify my stance.’ It seems enough of a concession to Kevin, who then as-
sumes the role of the main storyteller and continues the story.

4 K: he was trying to `sell us ^cologne
7 W: he was n’t trying to ^sell us cologne,
10 K: I guess he was trying to like,
11 ^lure us to a .. place
where they ^would sell,
12 like,
13 imitation cologne.
We can see from the diagraph that on line 7 Wendy is building her divergent
stance on Kevin’s just expressed stance. She does so with the negation and with
just slightly shifting the stress from cologne to sell. Kevin in turn builds his now
more converging stance on his own and Wendy’s immediately preceding turn,
by resonating with their syntactic form (he was trying to X us), and in addition,
bringing in some new information, a place and imitation cologne, that further
specify how he had perceived the original incident. The form ^would sell even
escalates the content in trying to sell, thus displaying that Kevin still maintains his
original position on this point.
In sum, I guess is brought in when the current speaker signals that, in re-
sponse to some evidence or other material presented in just prior turns, a (sud-
den) change in his or her state of knowledge or awareness has occurred and he or
she wishes to align and affiliate with the previous speaker and to express a (more)
convergent stance. At the risk of simplifying somewhat, the difference between I
think and I guess in second position is that I think tends to project or frame dis-
aligning or disaffiliative second actions, namely a different slant or take on a given
issue, or a new speaker perspective (Kärkkäinen 2003), whereas I guess basically
affiliates with the prior speaker. Even though very similar, the two are not inter-
changeable but have each specialized in a certain type of interactional function.
Lastly, I will briefly mention two other types of second-position actions
framed by I guess, in which the function of this frame comes close to the seman-
tic meaning of (weak) “belief ” as proposed by Chafe (1986) and Thompson and
Mulac (1991a). First, there is a typical reactive response to evaluative first actions,
Yeah, I guess so., where I guess marks the participant’s striving toward a conver-
gent stance (e.g., compliance). On occasion, it also takes on a sarcastic tone or
conveys joking reluctance in agreements with denigrating prior turns (A: I get a
little ahead of myself. B: <@> Yeah I guess you do </@>.; SBCSAE 0005). Second,
in answers to requests for information, the speaker may simply answer I ^guess or
preface his or her answer with I ^guess, and the function again comes close to the
semantic meaning of “belief.”
(8) (Car Sales SBCSAE, unpublished)

1 G: .. (H) They have `many uh,
2 .. ^Spanish `speaking `people go în there?
3 D: <YWN> U=h,
4 u=h,
5 I ^guess‑ </YWN>,
6 .. ^Small ^percentage,
7 .. ^Basically they’re .. ^Caucasians.
Here, I guess appears simply to express that the actual answer to be offered is
probably true.
4.3 I guess in extended turns: Inserting a stance
In many cases, I guess occurs in conversational sequences in which one main speak-
er produces a multi-unit turn or tells a story, while the other participants align as
recipients. Here, I guess projects an inserted stance that is often based on an infer-
ence by the teller; the stance may constitute a side sequence or an aside from the
main story line and its trajectory. Such an aside often works toward increasing un-
derstanding or appreciation of the telling or story. Here, I guess sometimes comes
very close to an actual hearsay evidential used to mark that the current speaker
positions him- or herself as not possessing or relating first-hand knowledge.
Here is an example from the same story as in example (6), about feeding the
Oscar with goldfish. The following extract comes just before the one in example
(6) and starts the orientation sequence of the story (the actual telling starts on
line 39).
(9) (Deadly Diseases SBCSAE 0015 <00:23:47>)

1 KEN: [Anyway,
2 I remember they had a,
3 (H) they had this fish in there called] an Oscar.
4 (H)= And they are these like,
5 .. [they are these really,
6 JOANNE: [<X> I remember- </X>,
7 yeah th- they’re big].
8 KEN: .. They’re these] big gnarly suckers.
9 They’re like [2um,
10 JOANNE: [2Yeah,
11 I remember what they look2] like.
12 KEN: .. (TSK) (H) They’re like m-2] --
13 f- --
14 I don’t know,
15 [six or eight inch]es,
16 JOANNE: [big eyes],
17 [2(H)2]
18 KEN: [2(H)2] and they’ve got,
19 ... they ònly eat like other <HI> ^fish </HI>.
20 They eat like small [^fish.
21 JOANNE: [(SNIFF)=]
22 KEN: And they eat like],
23 ... you know like ^goldfish.
24 <A> And I guess #the </A> goldfish or ^g=uppies,
25 get the `brunt of êverything.
26 .. (H) Poor ^guys.
27 (H)(Hx) They’re `like the,
28 you know,
29 they’re like the ^cattle of the @ànimal @^world.
30 [They’re br=ed only to f=eed other things,

31 JOANNE: [Yeah,
32 f=eeds.
33 Yeah,
34 they’re feeders.
35 KEN: (H) Anyway,
36 (H) anyway so,
37 JOANNE: They’re nothing but tanks of feeders,
38 in every pet shop (Hx)].
39 KEN: (H) I remember w-] watching them feed the Oscar once,
40 when I was a kid.
41 ... (TSK) And they dropped in the goldfish,
Throughout the orientation sequence Ken casts a very negative stance toward the
Oscars by using negatively evaluative lexemes (big gnarly suckers). In line 19, he
yet again evokes a negative stance toward them through only; not only are the Os-
cars big and ugly, but they are also predatory fish. From here onwards, he engages
in a kind of incremental turn design and escalating stancetaking activity. He first
asserts that Oscars eat small fish, and then establishes that goldfish are among
them. The discourse markers you know and like as well as rather strong primary
accent on ^goldfish can be taken as displays of his awareness that in just prior
discourse Joanne has told a story about a snake that was fed with goldfish (“this is
the type of fish that we talked about before”).
On lines 24–25, Ken makes a generalizing evaluative claim about the hard des-
tiny of goldfish, <A> And I guess #the </A> goldfish or ^g=uppies, get the `brunt of
êverything. (cf. Scheibman this volume). He is struck by the likeness of what he
is about to tell to what Joanne had just told before, and this is reflected in the way
this generalization is produced, namely with a “discovery” prosody: it is latched
on with and without a break and with a very fast tempo at the beginning of the IU,
it is generally very low in pitch, and the extreme case formulation êverything re-
ceives a rather strong primary stress (cf. Pomerantz 1986). The utterance can now
be heard as really making an inference on the basis of the previous story and the
present story-so-far, very likely to intensify interest in the upcoming story about
the dramatic feeding of the Oscar.
Ken then stops to evaluate the goldfish some more: he offers a token of sym-
pathy toward them, Poor ^guys, and further evaluates their lot in the animal
world (lines 27–30). It is possible and even likely that by engaging in such escalat-
ing stancetaking activity that gradually also becomes more specific, Ken wishes
to involve the story recipients in taking a stance on the same object, the goldfish,
at least to express their validation or ratification but possibly also a convergent
stance. But the recipients may not have been able to do so at line 26 (notice the
short pause), as ‘getting the brunt of everything’ is still rather vague in reference
and, what is more, the recipients have not yet heard Ken’s actual story. Once Ken
has elaborated his point some more, Joanne expresses a strong convergent stance
at the earliest possible point on lines 31–34, and Ken continues his story (Anyway,
anyway so).
The role of I guess in this conversational story is to display and project that the
speaker is about to digress from the main story line and insert a stanced comment
of some kind. The speaker opens this digression or side sequence with I guess
(or in this case And I guess) and closes it with anyway (cf. again Jefferson 1972
for side sequences initiated by a recipient and Lenk 1998: 71–78 for anyway after
general conversational digressions).12 Such initial and final marking or bracket-
ing of discourse sequences, of course, strengthens the interpretation of I guess as
a prospective discourse marker for which more support will be given in Section
5. The inserted stance, the generalizing inference, is of the same “recently discov-
ered” quality as we have seen in earlier sections of this paper. The recency here
arises from the speaker recognizing the similarity of import of the prior story as
well as the present story-so-far (or really upcoming story).
The next example briefly illustrates a case in which no visible uptake follows
from the recipient, but in which the turn design is nevertheless a very interactive
one. This is a story told by a young girl who is herself learning equine science and
in this story expresses her appreciation of a girl ferrier in Minnesota.
(10) (Actual Blacksmithing SBCSAE 0001 <00:11:18>)
1 LYNNE: (H) Anyway,

2 I was gonna tell you about that %- --
3 ... the girl,
4 you know,
5 there isn’t mary- .. very many,
6 .. ferrier girl=s you know?
7 LENORE: Uhhuh.
8 LYNNE: That do this?
9 (H) This summer I met one.
10 (H) Jorgensen’s,
11 they have ... one .. shoer,
12 that comes to their house,
13 all the time,
14 they’ve had him for years and years and years.
15 (H) <A> And I guess like in `Minnesota it’s real </A>
<LO> ^we=t </LO> ,
16 (0.7) and ` stuff you know?

17 So like,
18 .. they really have to watch their shoes,
19 for not coming off,
20 because the hoof wall,
21 is so much softer.
The story is told in Montana. The reference to Minnesota is explained by the fact
that Lynne has spent most of the summer there with her boyfriend and his fam-
ily, the Jorgensens. The protagonist of Lynne’s story is a girl ferrier, who, as it
turns out much later in the story, is incredibly strong and competent after only
nine months of ferrier college. Yet the inserted assessment in lines 15-16 acts as a
preface to a longer assessment sequence that is consequential for the understand-
ing of the story about the ferrier girl: the particular challenges that ferriers face
in Minnesota is that the horses’ hooves are too wet (because of the wetness of the
ground).
I guess is here used as a stance frame to project that such an assessment and in
fact a longer sequence elaborating it will follow. The assessment, produced some-
what faster than prior speech, is designed to continue even after the high rising
intonation at the end of the IU on line 16: there is no pause at all after this turn-
construction unit, and lines 15–16 are clearly pragmatically incomplete, requiring
elaboration (Ford and Thompson 1996). I guess is in large part doing organiza-
tional work in the telling, as it comes at a transition point from the main story
line, i.e., at the beginning of a side sequence (see also example 11 below). Again,
this digression is closed with anyway (see Lenk 1998: 65–68, 93–94 for anyway af-
ter digressions supplying relevant background information and but anyway used
for continuing the narration), even though this time the resumption of the main
story line about the girl ferrier comes considerably later after 46 lines:
67 LYNNE: (H) But anyway,

68 ... (TSK) they always have this one horseshoer,
69 and,
70 (H) this girl’s been .. in this f=errier,
71 ... uh,
72 she’s going to an actual .. ferrier college
.
Again by way of summary, while they are producing multi-unit turns such as
conversational stories, speakers mark through I guess that they are diverging from
the trajectory of the story or telling, to insert a stanced utterance or even a longer
sequence of such utterances. These involve taking an explicit stance toward some-
thing that is consequential for the story, arises from the story as a conclusion, or
serves as an explanation for the recipients to better appreciate the story, but is
often something that the narrator him- or herself just inferred from the story-so-
far. Similarly to first and second position I guesses, a shift in the epistemological
stance of the speaker is thus again evident.13 I guess may simultaneously mark
that the teller does not possess first-hand knowledge or was not the actual ex-
periencer of an event, but that the telling at this point is based on hearsay or on
some indirect evidence, and is also potentially disputable. In all, we can say that
I guess again organizes interaction in that it is used to draw the recipients’ atten-
tion to the current speaker’s stance, possibly also to invite them to appreciate and
ratify this stance. However, it is often also enough for recipients to align as pas-
sive recipients of an extended turn such as a conversational story, or display only
minimal ratification (or ratification through embodied action only). And finally,
it is worth noting that the I guess-framed stanced utterances or sequences deviate
from the original trajectory of talk, that is, stancetaking assumes a central role in
interaction.
5. I guess as a discourse marker
In this paper the view has been adopted (see Section 3.2) that I guess is one of
a group of epistemic/evidential/evaluative reusable fragments that speakers may
use as practices of turn construction (Thompson 2002). Such a view of grammar
is a novel one and emphasizes grammar as consisting of a “collection of crystal-
lizations of linguistic routines” used in the service of social interaction (Ford et
al. 2003: 120). However, this view fits in rather nicely with that proposed in a long
body of linguistic research that views tokens like I guess under the category of
‘discourse markers.’ As was established in Section 3.1 above, I guess is one of the
very frequent stanced items in spoken American English, even though not quite
as frequent as I think and I don’t know. While the latter have been shown in recent
research to have specialized into some rather routine discourse-organizing func-
tions in conversation (Beach and Metzger 1997; Kärkkäinen 2003; Scheibman
2000), we have also seen evidence in the examples above of a similarly discourse-
organizing function of I guess, as it may signal that an upcoming sequence is not
fully coherent with the main flow of the telling or talk (cf. Lenk 1998 for a view
of discourse markers as primarily marking global coherence relations). In this
section, I will relate the present findings to this body of research: this fragment
or token very largely fulfils the criteria proposed by Schiffrin in her seminal work
(1987: 328) for a linguistic item to count as a discourse marker (cf. an essentially
similar list by Brinton represented in Jucker and Ziv 1998: 3, and a summary of
criteria by different scholars in Helt 1997: 16–17). However, because the focus

in the present article has been on the role of I guess in the stancetaking activity
between conversational co-participants, and not on all possible environments of
use of this marker, some of the observations that I will make below of its discourse
marker status are not directly observable in the example segments above.
First, I guess is syntactically detachable from sentences, as it has largely lost
all overt marking of a syntactic connection to them, namely the complementizer
that. It also has some discourse and sentence mobility, because even though it
tends strongly to occur intonation-unit-initially and at the same time clause- and
sentence-initially (as in all the examples of this paper), it may also at times be en-
coded as an independent intonational entity, a separate IU, and may appear (even
though rarely) in the middle of an emergent syntactic sentence (for lack of space,
I do not present such occurrences here).
Schiffrin’s (1987) second criterion for discourse markers is that they have to
be commonly used in initial position of an utterance, which was established im-
mediately above.
Third, I guess has a range of prosodic contours, as it is produced with second-
ary, primary, or no stress on guess (see examples (3), (5)–(7), and (11) for second-
ary stress; (4) and (8) for primary stress; and (9) for an unstressed case), while at
times it is produced in a very accelerated and phonologically reduced form (ex-
amples (9) and (10)) or with parenthetical prosody (example (5)). Givón indeed
observes of “epistemic quantifiers” that “the conventionalized subject pronoun is
so specific to particular verbs, that it is often dropped in rapid speech,” as in (I)
think she’s there, or (I) guess you were right (1993: 38). However, only one such
instance was found in the present database, Doris: … Guess I’ll put it o=n. it’s not
on, <Actual Blacksmithing SBCSAE 0001 00:12:46>.
Fourth, I guess operates at both local and global levels of discourse, because
it can project an inserted stanced digression not only one utterance long (e.g.,
examples (5), (6), and (9)), but also a side sequence of several units (as in example
(10), where the side sequence extends over some 50 IUs).
Fifth, according to Schiffrin (1987) a discourse marker either has to have no
meaning, a vague meaning, or to be reflexive (of the language, of the speaker).
This criterion is also clearly filled by I guess. In many contexts of use, it seems
to make only a very general or vague reference to the actual speaker: it primar-
ily frames and projects an upcoming stancetaking action and simultaneously the
onset of a digression, a side sequence. But on other occasions, I guess appears to
retain some of its referential, lexical meaning and brings in a particular kind of
evidential stance, e.g., in example (4), the speaker bases his inference on visual
evidence of the masks on the wall (and stresses ^guess); or in example (10), the
teller does not have first-hand knowledge or experience about Minnesota. In the
latter cases, I guess acts much like a hearsay evidential such as people say or I’m
told, and displays the speaker’s fine-tuned positioning toward the knowledge that
she is offering (see Chafe 1986 on how some evidentials may be “borrowed” from
one function or mode of knowing to another, e.g., it seems was originally a marker
of induction but may sometimes be used as a hearsay evidential).
Of course we can argue that there is some semantic meaning to I guess, as
speakers do not simply choose I think or some other discourse marker in its
place – in other words, we cannot say that I guess has no meaning at all but a la-
tent (vague) meaning. The following two examples further display the difference
of I guess and I think as discourse markers. In example (11), Rebecca, an attorney,
is engaged in the delicate issue of preparing Rickie, the victim of an exhibitionist,
to appear as a witness in court. Here, Rickie is relating to Rebecca where exactly
on the train the defendant was when the incident took place: where he got in and
where he was seated (they have the layout of the train in front of them).
(11) (Tell the Jury That SBCSAE 0008 <00:08:11>)

1 RICKIE: He went ^through <X> th[ere </X>,
2 REBECCA: [He `went through ^these `doors]?
3 RICKIE: into ìt unhunh],
4 into ànother ^car.
5 REBECCA: Ôh[=].
6 RICKIE: [(H)] And `then he came `back âgain,
7 .. and `then,
8 <A> (H) I `guess that was `like another ^stop,
9 and more `people were getting ôff </A>,
10 .. and ^the=n,
11 ... that’s when he `came .. and ^s=at b=y me,
12 ^sat .. in this `seat right ^here.
We can see that lines 8–9 are inserted in a faster tempo to bring in information
that is subsidiary to the main activity of establishing the exact location of the de-
fendant. I guess inserts the speaker’s conjecture as to what happened otherwise on
the train when the defendant moved closer to her: Rickie infers that it was easier
for him to move because people were moving about anyway as they were getting
off the train (or, possibly, because there were now fewer people). Again the aside
contains the speaker’s speculation about what may have caused the man to come
back to her car, and the role of I guess is to mark it as such. Let us compare this
example to an almost similar one that came a little earlier in the same conversa-
tion, where we have I think instead of I guess, and the special nature of each will
become more apparent.
(12) (Tell the Jury That SBCSAE 0008 <00:06:53>)14

1 REBECCA: A=nd,
2 .. when he came in,
3 .. when .. where was he when you first saw him .. come in.
4 RICKIE: `He got on the ^trai=n ò=n like,
5 (THROAT) I ^think ^Twelfth Street or something?
6 in .. (H) Ôakland?
7 (H)<VOX> And like in the,
8 .. let me see </VOX>.
9 (3.9)
10 Okay.
11 .. He got on another car.
This is similarly a case of subsidiary information inserted in a lengthy answering

turn, as Rebecca’s question on lines 1–3 was really meant to establish where on
the train, rather than where in Oakland, the defendant had been when Rickie
first laid eyes on him. Rickie goes on to provide more information about whether
the defendant got in the same car or not on line 7. By inserting I think (and or
something) she simply marks part of the subsidiary information as uncertain. The
difference between the two markers is in the potentiality inherent in I guess to
display an inference, that is, to display that the speaker is realizing or inferring
something at the moment of speaking, often on the basis of available evidence.
The speaker therefore orients toward making a conclusion rather than simply of-
fering uncertain information. By contrast, I think marks speaker uncertainty and
orients more toward the speaker’s memory as the basis of uncertainty.
6. Conclusion
In this paper, I have argued that I guess, a reusable evidential (and not really epis-
temic) fragment, is frequently employed in American English conversation as a
stance frame, which may appear in rather diverse sequential positions in interac-
tion. My aim was to examine what kind of stance is expressed by this marker,
and what its role is in the stancetaking activity between conversational co-par-
ticipants. Despite the different sequential positions established, namely sequence-
initiating actions, responsive actions, and extended tellings or multi-unit turns,
the observed commonalities can be summarized as follows.
From a subjective, single-speaker vantage point:

– The form and decontextualized meaning of I guess is that of a subjective evi-
dential marker. The evidence in question is provided by prior discourse or
by visual or sensory evidence, and also via inference (i.e., with no overt evi-
dence in sight). Yet the actual referential meaning of this evidential is not very
strong in many instances of use, nor is it easy to determine without recourse
to the actual context of use.
– I guess projects a certain type of stance to be expressed in the utterance to fol-
low.
– Such a stance can variously be said to involve “making an inference,” “coming
to a realization,” or “drawing a conclusion,” i.e., I guess indexes a reasoning
or inferential process of the speaker, which involves a shift in the speaker’s
epistemological stance.

From an intersubjective perspective:
– The stance projected by I guess is not carried around ready-made but “surfac-
es” in the sequentially constructed intersubjective weave of dialogue. It arises
from, or is based on, evidence or some stimulus gleaned by the speaker (sto-
ryteller) from the ongoing or just prior turn(s), from the actual physical en-
vironment, or from the wider social context. I guess takes into account these
interactional contingencies and displays the recency of the speaker’s inference
and displayed stance.
– Depending on the sequential position, I guess-framed turns and actions may
function to invite others to take a stance, to display that the current speaker
gives up or modifies his or her stance (and not only aligns but also often af-
filiates with the recipient), and to draw the recipients’ attention to the current
speaker’s inserted stance (to increase and facilitate their understanding and
appreciation of a telling).
– Within these actions and turns-at-talk, I guess can be viewed as an organizer
of the stancetaking activity between co-participants. It is used to signal that
by making public a “just discovered” subjective stance, the speaker invites
others to also take a stance, gives up on or adopts a new stance, or alerts
recipients to an inserted stance at crucial points in a telling and invites them
to join in the stancetaking activity. The relatively high frequency of I guess in
conversational data serves as further proof for this organizing function: this
fragment can be fully regarded as a discourse marker that signals an upcom-
ing stanced digression or side sequence to follow.
– Finally, I guess appears at points in the interaction at which stance and stan-
cetaking is highlighted in favor of proceeding with business as usual; in that
sense, we can say that stance impinges on the projected trajectory of talk and
thus organizes interaction on a par with other interactional organizations

such as coherence, turn-taking, sequential organization, preference organiza-
tion, and repair organization.
As we saw in Section 3.2, Thompson (2002) argues that fragments like I guess pro-
vide a certain type of perspective or stance toward the actions (i.e., assessments,
claims, counterclaims, and proposals) being done in the associated utterance. In
light of my findings, however, it seems that I guess not so much provides a per-
spective toward the actions, as it signals and displays a just discovered stance to
the co-participants. The frame I guess is an essential part of the action being ac-
complished in the turn. An action, such as an assessment or a claim, can be well
accomplished without I guess, but it would then constitute an essentially different
type of assessment or claim (cf. example (2) above: I guess that means his bro-
ken leg is @doing @okay versus That means his broken leg is doing okay). In other
words, I guess signals, projects, and makes more explicit the kind of stancetaking
action that is being done in the current turn.
We have seen, on one hand, that stance marking can be of a very routin-
ized nature linguistically, in that speakers only use a rather small set of inher-
ently stanced words with some frequency. If, however, we turn our focus to stan-
cetaking as a process and an activity oriented to by participants, we can see that
some of these frequent markers then develop routine functions as organizers of
such stancetaking activity and do not in themselves express a clear, unambiguous
“stance” anymore. Finally, if we approach stance as something jointly oriented to
by the co-participants, we need even more spread in our linguistic description.
As I hope to have shown in the above analysis, some linguistic practices of stan-
cetaking go beyond specific, discrete grammatical or lexical devices analyzable
in single speakers’ contributions. I have demonstrated some syntactic, semantic,
and prosodic resonances between contributions by different speakers (especially
when I guess occurs in second position), which are also to be seen as resources for
stancetaking and therefore deserve our full attention.
Notes
1. This paper, as also those by Pentti Haddington, Tiina Keisanen, and Mirka Rauniomaa, is
based on work done in my research project entitled Interactional Practices and Linguistic Re-
sources of Stance Taking in Spoken English (2002–2006), and has been financed by the Academy
of Finland (grants 00381 and 53671). The project is greatly indebted to our collaborator John
Du Bois, who has provided us with a view of stance as achieved out in the social world and in
interactions with other social agents. My heartfelt thanks are also due to Robert Englebretson
for making the 10th Biennial Rice University Linguistics Symposium happen, as well as for
many invaluable comments on earlier versions of this paper. Thanks also to all the participants
of the Rice symposium for many helpful comments on my paper, and to the two anonymous
referees for making many important points and helpful suggestions for revisions. I myself am of
course responsible for what use I have made of the comments and suggestions I have received,
as also for all the remaining inadequacies.
2. This finding is nicely corroborated by Precht’s work within the “appraisal” framework (cf.
Martin 2000, 2003):
The resources of language enable a virtually unlimited number of ways in which

we could express ourselves, however, my results suggest that we are culturally ‘pro-
grammed’ to use a very limited, very specific subset of these options. My previous
analysis found more than 1,400 different stanced words in English, and yet we use
only about 150 words for ninety percent of our stance expression [..]. Although we have
a myriad of options for expressing our emotions, attitudes, and commitments, we
tend to use the same small set of stance markers repeatedly. Our expression of stance,
I would argue, is shaped by culture and custom – we are socialized to use particular
stance markers in particular ways. (Precht 2003: 240, emphases mine)
3. This extremely high frequency is said to be largely due to the use of the clause I think.
4. The frequency of the verb (or noun) guess turned out to be only very little higher than that
for I guess (23). By contrast, the total frequencies for the verbs think and know (1,115 and 1,713
respectively) were considerably higher than the ones for I think and I don’t know (729 and 309).
Upon closer inspection these high figures are mainly accounted for by the inclusion of other
highly grammaticized discourse markers (in addition to I think and I don’t know), namely I
don’t think (118), you know (954), and I know (224).
5. The ICE-GB data are essentially collected for corpus linguistics purposes, i.e., to form a
computer-searchable syntactically tagged and parsed database, with less attention to transcrip-
tion detail or quality of sound, and it is in many parts not ideally suited for close interactional
analysis.
6. But see for example Quirk et al. (1985: 1112–1113) who hold the opposite view, i.e., consid-
ering “comment clauses” like I think to be subordinate to the rest of the sentence.
7. Similar observations about the frequency and formulaic nature of such explicitly subjective
collocations have further been made of many other languages. Thus, Weber and Bentivoglio
(1991) discuss essentially similar discourse patterns of the Spanish verbs of cognition, creer ‘be-
lieve’ and pensar ‘think,’ in spoken data. For Swedish, Dahl (2000) reports on the high frequency
of first-person pronouns (but also second-person pronouns and generic pronouns) clustering
with mental verbs like tro ‘believe,’ tycka ‘think,’ tänka ‘think,’ minnas ‘remember,’ etc., and in
a recent study Karlsson (2003) discusses the interactional uses of the epistemic stance marker
jag tycker/tycker jag ‘I think’ in spoken Swedish data. First-person subjects typically co-occur
with mental verbs also in colloquial Finnish (mä luulen ‘I believe/think’) and Estonian (ma
arvan ‘I think,’ mai tea ‘I don’t know’; Keevallik 2003), even though one of the frequent Finnish
epistemic stance markers, minun mielestä ‘in my opinion’ and minusta ‘I think,’ only displays a
first-person subject but no verb (Rauniomaa this volume).
8. I guess is often translatable with just a particle to other languages, e.g., with the Finnish kai
or the Swedish väl, both meaning ‘probably, maybe.’ This could be taken as further evidence for
its discourse marker status.
9. Where evidentiality fits in with epistemicity and which one is considered the superordinate
category varies from one researcher to the next. Chafe (1986), in fact, subsumes epistemic mo-
dality under evidentiality, or discusses evidentiality as coding both the speaker’s attitude toward
the reliability of knowledge and his or her source of knowledge or mode of knowing, whereas
Biber et al. (1999) include under epistemic stance not just markers of certainty, actuality, preci-
sion, and limitation but also of source of knowledge and of the perspective from which the
information is given.
10. I thank John Du Bois for the observation that it is possible to describe the use of I guess
as “I’m discovering something about my own subjectivity as I speak.” Subjectivity in this case
comes out of and presupposes intersubjectivity, or the speakers engaging with other subjectivi-
ties in conversational dialogue (see also Du Bois this volume).
11. This way of framing the upcoming story is very similar to a more explicitly evaluative device
for prefacing stories, namely by what Goodwin (1996) has called “prospective indexicals.” Items
like problem or a wonderful/terrible thing in a story preface offer a framework for interpretation
to the recipients, who do not yet have access to the story but who are expected to respond to it
in an appropriate way upon its completion.
12. I thank Robert Englebretson for drawing my attention to this, in fact for three of my ex-
amples (6), (9), and (10).
13. I thank the anonymous reviewer for pointing out this obvious similarity to me.
14. This example is analyzed in more detail in Kärkkäinen (2003).
References
Anward, J. 2000. “How types of lexical resources emerge from turn-construction.” Paper pre-
sented at the Euroconference on Interactional Linguistics, Spa, Belgium, September 16–
21.
Beach, W.A. and Metzger, T.R. 1997. “Claiming insufficient knowledge.” Human Communica-
tion Research 23: 560–585.
Biber, D., Johansson, S., Leech, G., Finegan, E. and Conrad, S. 1999. The Longman Grammar of
Clift, R. 2006. “Indexing stance: Reported speech as an interactional evidential.” Journal of So-
ciolinguistics 10 (5): 569–595.
Chafe, W. 1986. “Evidentiality in English conversation and academic writing.” In Evidentiality:
The Linguistic Coding of Epistemology, W. Chafe and J. Nichols (eds.), 261–272. Norwood,
NJ: Ablex Publishing Corporation.
Dahl, Ö. 2000. “Egophoricity in discourse and syntax.” Functions of Language 7 (1): 37–77.
Downing, A. 2001. “‘Surely you knew!’ Surely as a marker of evidentiality and stance.” Functions
of Language 8 (2): 251–282.
Du Bois, J.W. 2003. “Stance and consequence in interaction.” Paper given at the Langnet Sym-
posium, University of Oulu, September 11.
Du Bois, J.W., Schuetze-Coburn, S., Cumming, S. and Paolino, D. 1993. “An outline of dis-
course transcription.” In Talking data: Transcription and Coding in Discourse Research, J.
Edwards and M. Lampert (eds.), 45–87. Hillsdale, NJ: Lawrence Erlbaum.
Du Bois, J.W., Chafe, W.L., Meyer, C. and Thompson, S.A. 2000. Santa Barbara Corpus of Spo-
ken American English, CD, Part 1. Philadelphia: Linguistic Data Consortium.
Du Bois, J.W., Chafe, W.L., Meyer, C., Thompson, S.A. and Martey, N. 2003. Santa Barbara Cor-
pus of Spoken American English, CD, Part 2. Philadelphia: Linguistic Data Consortium.
Duvallon, O. and Routarinne, S. 2005. ”Parenthesis as a resource in the grammar of conversa-
tion.” In Syntax and Lexis in Conversation. Studies on the use of linguistic resources in talk-
in-interaction, A. Hakulinen and M. Selting (eds.), 45–74. Amsterdam: John Benjamins.
Ford, C. and Thompson, S.A. 1996. “Interactional units in conversation: syntactic, intonational,
and pragmatic resources for the management of turns.” In Interaction and Grammar, E.
Ochs, E. Schegloff and S.A. Thompson (eds.), 134–184. Cambridge: Cambridge University
Press.
Ford, C., Fox, B. and Thompson, S.A. 2002. “Social interaction and grammar.” In The New Psy-
chology of Language: Cognitive and Functional Approaches to Language Structure, Vol. 2, M.
Tomasello (ed), 119–143. Mahwah, NJ: Lawrence Erlbaum.
Givón, T. 1993. English Grammar. A Function-Based Introduction. Vol. 1-2. Amsterdam: John
Benjamins.
Goodwin, C. 1981. Conversational Organization. Interaction between Speakers and Hearers.
New York: Academic Press.
Goodwin, C. 1996. “Transparent vision.” In Interaction and grammar, E. Ochs, E. Schegloff and
S.A. Thompson (eds.), 370–404. Cambridge: Cambridge University Press.
Goodwin, C. and Goodwin, M.H. 1992. “Assessment and the construction of context.” In Re-
Haddington, P. 2003. Oulu Corpus of US / British Television Interviews from 1999–2004. Ms.,
Department of English, University of Oulu.
Helt, M.E. 1997. Discourse Marker and Stance Adverbial Variation in Spoken American English:
A Corpus-Based Analysis. Ph.D. dissertation. Department of English, Northern Arizona
University.
Heritage, J. 1984. “A change-of-state token and aspects of its sequential placement.” In Struc-
tures of Social Action: Studies in Conversation Analysis, J.M. Atkinson and J. Heritage (eds.),
299–345. Cambridge: Cambridge University Press.
Heritage, J. 2005. “Cognition in discourse.” In Conversation and Cognition, H. te Molder and J.
Potter (eds.), 184–202. Cambridge: Cambridge University Press.
Holquist, M. 1990. Dialogism. Bakhtin and His World. London: Routledge.
International Corpus of English–Great Britain (ICE-GB). Survey of English Usage, University
College London.
Jefferson, G. 1972. “Side sequences.” In Studies in Social Interaction, D. Sudnow (ed.), 294–338.
New York: Free Press.
Jefferson, G. 1979. “A technique for inviting laughter and its subsequent acceptance/declina-
tion.” In Everyday language: Studies in ethnomethodology, G. Psathas (ed.), 79–96. New
York, NY: Irvington Publishers.
Jucker, A. and Ziv, Y. 1998. “Discourse markers: Introduction.” In Discourse Markers: Descrip-
tions and Theory, A. Jucker and Y. Ziv (eds.), 1–12. Amsterdam: John Benjamins.
Kärkkäinen, E. 2006. “Stance taking in conversation: From subjectivity to intersubjectivity.”
Text & Talk 26 (6): 699–731.
Karlsson, S. 2003. “Interactional uses for an epistemic marker: The case of ‘jag tycker’/‘tycker
jag’ in Swedish.” Melbourne Papers in Linguistics and Applied Linguistics 1: 5–23.
Keevallik, L. 2003. From Interaction to Grammar. Estonian Finite Verb Forms in Conversation.
Acta Universitatis Upsaliensis. Studia Uralica Upsaliensia 34.
Kendon, A. 2004. Gesture: Visible Action as Utterance. Cambridge: Cambridge University
Press.
Lenk, U. 1998. Marking Discourse Coherence. Functions of Discourse Markers in Spoken English.
Tübingen: Gunter Narr.
Lerner, G. 1991. “On the syntax of sentences-in-progress.” Language in Society 20 (3): 441–
458.
Lerner, G. 1992. “Assisted storytelling: Deploying shared knowledge as a practical matter.”
Qualitative Sociology 15 (3): 247–271.
Lerner, G. 1996. “Finding ‘face’ in the preference structures of talk-in-interaction.” Social Psy-
chological Quarterly 59 (4): 303–321.
Local, J. 1992. “Continuing and restarting.” In The Contextualization of Language, P. Auer and
A. di Luzio (eds.), 273–296. Amsterdam: John Benjamins.
Mandelbaum, J. 1989. “Interpersonal activities in conversational storytelling.” Western Journal
of Speech Communication 53: 114–126.
Mandelbaum, J. 1993. “Assigning responsibility in conversational storytelling.” Text 13 (2):
247–266.
Martin, J.R. 2000. “Beyond exchange: APPRAISAL systems in English.” In Evaluation in Text.
Martin, J.R. 2003. “Introduction.” Text. Special issue: Negotiating Heteroglossia: Social Perspec-
tives on Evaluation, 23 (2): 171–181.
ity, J. Gumperz and S. Levinson (eds.), 407–437. Cambridge: Cambridge University Press.
Ochs, E. and Schieffelin, B. 1989. “Language has a heart.” Text 9 (1): 7–25.
Pomerantz, A. 1984. “Agreeing and disagreeing with assessments: Some features of preferred/
dispreferred turn shapes.” In Structures of Social Action. Studies in Conversation Analysis,
J.M. Atkinson and J. Heritage (eds.), 57–101. Cambridge: Cambridge University Press.
Pomerantz, A. 1986. “Extreme case formulations.” Human Studies 9: 219–230.
Potter, J. 1998. “Discursive social psychology: From attitudes to evaluations.” European Review
of Social Psychology 9: 233–266.
Potter, J. and Hepburn, A. 2003. “‘I’m a bit concerned’ – early actions and psychological con-
structions in a child protection helpline.” Research on Language and Social Interaction 36
(3): 197–240.
Precht, K. 2003. “Stance moods in spoken English: Evidentiality and affect in British and
American conversation.” Text. Special issue: Negotiating Heteroglossia: Social Perspectives
on Evaluation 23 (2): 239–257.
Schegloff, E. 1996. “Turn organization: On the intersection of grammar and interaction.” In
Interaction and Grammar, E. Ochs, E. Schegloff and S.A. Thompson (eds.), 52–133. Cam-
bridge: Cambridge University Press.
Scheibman, J. 2000. “I dunno ... A usage-based account of the phonological reduction of don’t in
American English conversation.” Journal of Pragmatics 32: 105–124.
Scheibman, J. 2001. “Local patterns of subjectivity in person and verb type in American English
conversation.” In Frequency and the Emergence of Linguistic Structure, J. Bybee and P. Hop-
per (eds.), 61–89. Amsterdam: John Benjamins.
Schiffrin, D. 1987. Discourse Markers. Cambridge: Cambridge University Press.
Thompson, S.A. 2002. “‘Object complements’ and conversation: Towards a realistic account.”
Studies in Language 26 (1): 125–163.
Thompson, S.A. and Mulac, A. 1991a. “A quantitative perspective on the grammaticization of
epistemic parentheticals in English.” In Approaches to Grammaticalization. Vol. 1–2, E.C.
Traugott and B. Heine (eds.), 313–329. Amsterdam: John Benjamins.
Thompson, S.A. and Mulac, A. 1991b. “The discourse conditions for the use of the complemen-
tizer that in conversational English.” Journal of Pragmatics 15: 237–251.
Traugott, E.C. 1995. “Subjectification in grammaticalisation.” In Subjectivity and Subjectivisa-
tion: Linguistic Perspectives, D. Stein and S. Wright (eds.), 31–54. Cambridge: Cambridge
University Press.
Voloshinov, V.N. [1930] 1973. “Verbal interaction.” In Marxism and the Philosophy of Language,
Chap. 3, 83–98. Cambridge, MA: Harvard University Press.
Weber, E. and Bentivoglio, P. 1991. “Verbs of cognition in spoken Spanish: A discourse pro-
file.” In Discourse–Pragmatics and Verb. The Evidence from Romance, S. Fleischman and L.
Waugh (eds.), 194–213. London: Routledge.
Appendix: Symbols used in transcription
(From Du Bois et al. 1993)
Units
Intonation unit {carriage return}
Truncated word -
Transitional continuity
Final .
Continuing ,
Appeal (seeking a validating response
from listener) ?
Speakers
Speech overlap (numbers inside brackets
index overlaps) [ ]
Accent and lengthening

Primary accent (prominent pitch movement
carrying intonational meaning) ^
Secondary accent `
Unaccented
Lengthening =
Pause
Long and medium
(length indicated in seconds) …(1.0)
Short (brief break in speech rhythm) ..
Vocal noises
e.g., (TSK), (SNIFF), (YAWN), (DRINK)
Glottal stop %
Exhalation (Hx)
Inhalation (H)
Laughter (one pulse) @
Laughter during speech (1–5 words) @
(e.g. @two @words)
Laughter during speech (+6 words) @
(e.g. <@> words </@>)
Quality
Special voice quality <VOX>words</VOX>
Forte: loud <F>words</F>
Higher pitch level <HI>words</HI>
Lowered pitch level <LO>words</LO>
Parenthetical prosody <PAR>words</PAR>
Allegro: rapid speech <A>words</A>
Marcato: each word distinct and emphasized <MRC>words</MRC>
Yawning <YWN>words</YWN>
Transcriber’s perspective
Uncertain hearing <X> words </X>
Uncertain #word
Stance markers in spoken Finnish
Minun mielestä and minusta in assessments
Mirka Rauniomaa
University of Oulu, Finland
1. Introduction1
The expressions minun mielestä and minusta, which frequently occur in almost
any Finnish interaction, have drawn analysts’ attention before but mainly ended
up in side comments and footnotes. The purpose of this paper is to contribute to
the understanding of the two expressions by focusing solely on their locally situ-
ated use, particularly in the context of assessments. I argue that their use stems
from participants’ need to communicate how the explicitly evaluative action that
they are performing fits in the ongoing interaction. Indeed, the analysis is guided
by the idea that social interaction organizes linguistic phenomena, as posited in
interactional linguistics (see, e.g., Ford et al. 2002; Ochs et al. 1996; Selting and
Couper-Kuhlen 2001). The central theme pursued in this paper is the notion of
stancetaking, i.e., participants’ joint construction of evaluations, attitudes, affec-
tive stances, etc. in dialogic interaction (Du Bois this volume; Kärkkäinen 2006,
this volume). Before proceeding to a discussion of stancetaking, however, a closer
look at the expressions is in order. In what follows, I will discuss minun mielestä
and minusta in terms of their morphosyntactic form and with relation to previous
analyses. I will also briefly introduce the data used in this study (Section 2).
The expression minun mielestä consists of the first-person-singular pronoun
minä in its genitive case and the noun mieli ‘mind’ in its elative, i.e., inner loca-
tive, case. The literal translation of minun mielestä would thus be ‘from my mind’
(or ‘out of my mind’); other translations into English include ‘in my opinion,’ ‘to
my mind,’ and ‘it seems to me.’ Similar meanings can be attributed to the expres-
sion minusta, which consists simply of the first-person-singular pronoun minä
and the elative case ending -stA (the capital A reflects Finnish vowel harmony:
depending on the vowels in the stem, the ending takes either the back vowel a or
222 Mirka Rauniomaa
the front vowel ä). Literally minusta would translate as ‘from me’ (or ‘out of me’);
the most appropriate translation is ‘I think.’ Both expressions are often reduced
phonetically, to the extent that minun mielestä may come close to sounding like
minusta, and minusta may take an even more reduced form. A comprehensive
analysis would certainly be useful, but I will not explore the phonetic realizations
of the expressions in any detail here. It is worth pointing out, however, that the
structural similarity of the phrases and the possible phonetic reduction of minun
mielestä suggest that minusta can be regarded as a contraction of minun mielestä.
For this reason, and because of data limitations, I do not attempt to distinguish
between them at this point. Nonetheless, whenever the analysis permits, I keep
the two expressions separate to encourage further investigations into their pos-
sibly differentiated use (see Luukka 1992a: 118–119, 139–140, 1992b: 373, for ob-
servations on the diverse distribution of the expressions in spoken and written
academic discourse).
In previous literature, minun mielestä and minusta have commonly been de-
scribed as hedges that are used to obscure the import of an utterance in order
to address issues of politeness, i.e., saving the speaker’s or a co-participant’s face
(Aalto 1997: 58–61; Hakulinen et al. 1989: 118–123; Lampinen 1990: 84–85; Luuk-
ka 1992a: 91–92, 108, 1992b: 363–373; Nikula 1996: 115–121, 135–136). More spe-
cifically, they have been considered to represent a particular type of hedge that
mitigates speakers’ commitment to the truth of a proposition (Hakulinen et al.
1989: 118–119; drawing on Prince et al. 1982: 85). In this way, the expressions de-
limit the generalizability of an utterance by framing it not as a fact but as an opin-
ion that the co-participants may disagree with (Luukka 1992a: 108, 1992b: 372–
373). Aalto (1997: 65) points out that whether or not such a delimitation in effect
strengthens or weakens the speaker’s commitment to a proposition is entirely
context-dependent. As an example, Aalto argues that a turn-initial minun mielestä
strengthens the expression of opinion and the speaker’s commitment to it when
it is placed in a turn in which the speaker gives grounds for his/her dissenting
opinion.
Overall, previous studies provide extensive overviews of linguistic modality
in general and, consequently, only touch upon such individual expressions as mi-
nun mielestä and minusta. What is more, they consider the two expressions to
mark speakers’ individual, subjective stances. The current paper builds on this
foundation to bring in the intersubjective aspects of stancetaking. I refrain from
the predetermined categories of politeness and carefully examine minun mielestä
and minusta in the specific sequential contexts in which they occur. In this way,
the present paper parallels Kärkkäinen’s (this volume) exploration of I guess and,
more importantly, her (Kärkkäinen 2003) study on I think, an expression that
is more or less equivalent to the Finnish minun mielestä and minusta (see also
Stance markers in spoken Finnish 223
Karlsson 2003 for a discussion of the Swedish jag tycker/tycker jag ‘I think’). Kärk-
käinen (2003) identifies several local purposes for the use of I think: depending on
its sequential positioning, I think marks topical and other boundaries, brings in a
personalized speaker perspective, displays on-line planning, signals turn comple-
tion or pursues a response from the recipients. As Section 5 will show, minun
mielestä and minusta can be seen to serve similar interactional functions.
The remarks that have been made on minun mielestä and minusta in previ-
ous literature suggest that the two items offer a routinized way for speakers to put
their stamp on an utterance. This paper proposes that the routinization is also
evident at a sequential level. The term stance marker best captures this nature of
the expressions: it implies that minun mielestä and minusta frame an utterance or
a turn as relevant in terms of stance. As access to knowledge plays a major part in
some cases in the present data, minun mielestä and minusta could be further cat-
egorized as epistemic stance markers. Nonetheless, the starting point of this paper
is not primarily on epistemicity per se but simply on answering the question why
minun mielestä and minusta are used in the context of assessments.
After a presentation of the data in Section 2, I will discuss stancetaking and
assessments in some detail in Section 3. I will then give a brief overview of the
stance markers’ positioning within an intonation unit and in the larger context of
a sequence (Section 4). The bulk of the paper, Section 5, presents the three inter-
actional functions of the stance markers that can be found in the data: projecting
disagreement in a second assessment, marking transition to a first assessment
across turns, and marking transition to a first assessment within an extended
turn. I will provide more evidence of the contingent nature of the stance markers
by presenting a special case in Section 6, before a discussion of the findings in the
conclusion (Section 7).
2. Data
The data are drawn from the corpus of conversational Finnish maintained by
the Department of Finnish Language and Literature at the University of Helsinki
(Keskusteluntutkimuksen arkisto). They amount to approximately two and a half
hours, or around 34,000 words, of casual conversation: there are 2 face-to-face
conversations, each lasting for about an hour, and 12 telephone conversations,
ranging in length between 1 and 14 minutes. The data were originally transcribed
using the transcription conventions developed by Gail Jefferson (as laid out in
Seppänen 1997), but for the present study, I have re-transcribed the relevant parts
according to the intonation-unit-based discourse transcription system (Du Bois
et al. 1992; Du Bois et al. 1993). In addition to a free English translation, I have
224 Mirka Rauniomaa
added a morpheme-by-morpheme gloss for each line of transcript. Keys to rel-

evant transcription and glossing symbols are provided in Appendices A and B.
It should be pointed out that the speakers in these data come mainly from
southern and eastern Finland, which represent two distinct dialect areas. This
explains the varied spellings of the stance markers under examination (e.g., mi-
nun mielestä ~ mun/miun mielest(ä); minusta ~ must(a)/miust(a)) and possibly of
some other words in the transcripts. In referring to the stance markers in general,
I use the full forms minun mielestä2 and minusta, but I deem it important that the
evident dialect differences be reflected in the transcripts.
All in all, 67 occurrences of the stance markers can be found in the data.
Minun mielestä is somewhat more frequent with a total of 42 cases, minusta be-
ing represented by a total of 25 cases. Examples (1) and (2) typify the expressions
minun mielestä and minusta, respectively.
(1) sg067_a2 Lawyer Issues <T:00:03:35>

R: Se on mu-n miele-stä tosi hyvä kirja.
it be:3sg I-gen mind-ela really good book
‘In my opinion it is a really good book.’
(2) sg151_a Chat about Something <T:00:37:33>

I2: Mut ol-i miu-sta ne hyvä-n-näköse-t.
but be-pst:3sg I-ela they good-gen-looking-pl
‘But I think they were good-looking.’
Although it is possible to use mielestä or simply the elative case ending -stA with
other person references, e.g. with another personal pronoun or with a proper
name, such cases are rare in comparison with the first-person-singular forms. In
the present data, there is only one case of a second-person reference (example 3)
and one case of a third-person reference, in which the speaker refers to a cat called
Perttu (example 4).

I2: Mut on-k-s nää nyt siu-n miele-st sit hyvä-t.
but be:3sg-q-cli these now you-gen mind-ela then good-pl
‘But are these ones good in your opinion.’

R: Se-n mie- Pertu-n miele-stä se on aina just ede-ssä,
it-gen cat_name-gen mind-ela it be:3sg always just front-ine
‘In his opi- in Perttu’s opinion it is always in his way,’
Examples (3) and (4) represent individual cases; the stance-marker elements
mielestä and -stA are overwhelmingly used with first-person reference in the cur-
rent data. This observation corresponds with Helasvuo’s (2001a: 32) finding that
first-person subjects typically co-occur with mental verbs, or verbs of cognition,
in spoken Finnish. Similar collocations between first-person referents and stance-
inferring expressions in subject-verb combinations have also been observed in
other languages (see Dahl 2000 for Swedish; Keevallik 2003: 74–100 for Estonian;
Kärkkäinen 2003; Scheibman 2001 for English; Weber and Bentivoglio 1991 for
Spanish). The Finnish minun mielestä and minusta, as adverbial noun phrases in-
cluding an explicit reference to cognition and/or the speaker, can also be seen as
indices of the strong tendency in interactive discourse for speakers to personalize
what they are saying, i.e., to talk from their particular point of view (see Bybee
and Hopper 2001). It has to be added, however, that even this personalization, or
subjectivity, can only be constructed in interaction with other participants. The
interactive nature of stancetaking will be discussed next.
3. Stancetaking through assessments
It is a well-established observation that when producing an utterance, speakers

leave in it traces of their attitudes, emotions, and so on, toward whatever they are
talking about (see Thompson and Hunston 2000 for a review). What is new is the
shift in focus from individual stances to stancetaking, i.e., from single-speaker
contributions to the joint construction of stance at both linguistic and interac-
tional levels (see Wu 2004: 3–19 for a review). Following this line of study, I exam-
ine the linguistic items that are the focus here, minun mielestä and minusta, within
their local sequential contexts and consider them to arise contingently from those
contexts. In other words, minun mielestä and minusta are here regarded to acquire
their status as stance markers only in and through talk-in-interaction.
The interactional nature of stancetaking is well spelled out in Du Bois’s (this
volume) definition of stance as being composed of three interlinked elements:
evaluating an object, positioning the self, and aligning with other subjects. Almost
any utterance helps to locate the speaker’s perspective or point of view among an
infinite set of possible ones. Within this view of stance, Du Bois emphasizes the
significance of the stance differential, a subtle difference in the participants’ align-
ment with each other, which gradually rises out of their contributions. To put it
differently, even if participants fully agree on a matter, their specific utterances are
affected by the fact that one precedes, follows, or overlaps the other.
Stancetaking is a social activity that is not restricted to any specific action
type. Nonetheless, certain action types can be considered to be particularly rele-
226 Mirka Rauniomaa
vant for stancetaking: assessments (as defined by Goodwin and Goodwin 1987: 9)
are explicitly evaluative and as such clearly contribute to stancetaking. In Finnish,
assessments typically follow more or less the same syntactic form as in English
(Tainio 1996: 85), a pattern that can be presented as follows (slightly modified
from Goodwin and Goodwin 1987: 22):
[third-person pronoun] + [copula] + [adverbial intensifier] + [assessment term]
As is evident in this syntactic pattern, assessments include an evaluation (assess-

ment term) of a particular referent (third-person pronoun) and simultaneously
mark the speaker’s position in relation to that referent and the co-participants,
e.g., by choice of tense in the copula (Goodwin and Goodwin 1992: 165). What is
more, assessments make relevant a type-connected uptake (Pomerantz 1984) and
thus facilitate participants’ alignment with each other. Assessments clearly rep-
resent the speaker’s perspective, and there is no need for an explicit reference to
the speaker. Nevertheless, as the following sections will show, minun mielestä and
minusta frequently occur within assessments, accomplishing orientational work.
4. Positions of minun mielestä and minusta
The way in which a single utterance is constructed is consequential to its interpre-

tation. It is therefore important to consider the immediate context of the stance
markers minun mielestä and minusta. I will here discuss two levels of positioning,
with a special focus on stance markers within assessments. Firstly, I will examine
where the stance markers are placed within the intonation unit in which they oc-
cur. Secondly, I will make a few remarks on the stance markers’ possible positions
in the sequence of actions, i.e., whether the stance markers occur in initiating or
responding turns.
The four possible positions that the stance markers can take in the intonation
unit (IU) are presented in Table 1. Table shows that there is a tendency for both
stance markers, especially minusta, to be placed at the start of an intonation unit.
Helasvuo (2001b: 145–149) has found that the clause core (i.e., the predicate and
its core arguments) is typically produced within one intonation unit in Finnish.
The frequent IU-initial positioning of stance markers thus reflects the fact that
they are an integral part of the utterance that they occur in, and that they provide
a frame for the whole upcoming clause or in some cases for a longer segment of
talk (see Hakulinen et al. 1989: 136). In the case of the English I think, Kärkkäinen
argues that its IU-initial positioning “helps recipients to align themselves to what
is coming” (2003: 115). In other words, stance markers orient participants to treat
Table 1. Position of stance markers minun mielestä, and minusta in the intonation unit
(assessments in parentheses)
Initial* Medial Final Separate Total
minun mielestä 24 (13) 15 (6) 1 (0) 2 (1) 42 (20)
minusta 20 (5) 5 (4) 25 (9)
Total 44 (18) 20 (10) 1 (0) 2 (1) 67 (29)
* Cases in which a stance marker is preceded by a discourse particle such as no ‘well’, niin ‘so’ and mutta
‘but’ are included within this group (see Hakulinen et al. 1989 for a discussion of discourse particles in
Finnish).
what follows as possibly requiring a response and in any case relevant to their
stancetaking. Additionally, if an IU-initial position coincides with a turn-initial
one, the stance marker can be considered to be especially relevant in making clear
the current turn’s relation to the previous one (see Sacks et al. 1974: 722).
When the stance markers are positioned somewhere in the middle of an in-
tonation unit, only a limited set of items can typically be placed in front.3 These
include the negation element ei,4 the verb on ‘is,’ pronouns, or a combination of
these. Their occurrence at that particular slot can be explained by the syntactic
constraints of spoken Finnish: the negation element is commonly placed at the
start of an intonation unit, and the favoured word order is SVO, with given in-
formation preceding new information in a clause and especially with pronominal
subjects preceding the verb (Helasvuo 2001b: 76–78). It is striking that in all the
cases in which a stance marker is placed IU-medially in an assessment, the evalu-
ative element comes only after the stance marker, either in the same intonation
unit or in the subsequent one. This suggests that IU-medial stance markers orient
participants to what follows in much the same way as IU-initial ones.
As Table 1 reveals, there is only one case of an IU-final stance marker in the
data. It occurs in the utterance no ei mun mielestä ‘well not in my opinion,’ which is
produced as an answer to a previous speaker’s question, and which is syntactically
tied to that question. The two cases of minun mielestä in which the stance marker is
placed in a separate intonation unit refer back to and cast new light on the speaker’s
previous utterance. Motivation for such use of the stance markers is also inextrica-
bly tied to the ongoing interaction, as example (6) will later demonstrate.
As for their possible positions in a somewhat larger context, both stance mark-
ers can be in turns that initiate new sequences or respond to a previous action.
Two types of action that can take place in either of these positions figure promi-
nently in the data: statements, or information delivery in a wider sense (Heritage
1984), and assessments, which typically come in pairs of first and second assess-
ments (Pomerantz 1984). In the present data, it is common for both stance mark-
ers to occur in assessments: 20 cases of minun mielestä (48%) and 9 of minusta
228 Mirka Rauniomaa
(36%) are produced within an utterance accomplishing such an action. That is to

say, the stance markers are frequently used in a context that can also otherwise be
identified as being about stancetaking.
These observations quite clearly indicate that the stance markers minun
mielestä and minusta are made relevant by the ongoing interaction. They do not
in themselves reveal speakers’ stances but point out how an utterance is to be un-
derstood there and then. This is the case especially when the stance markers occur
within assessments, i.e., when there are clearly evaluative elements either within
the intonation unit itself or in an adjacent intonation unit within the same turn. I
will examine such cases in detail in the following sections.
5. Minun mielestä and minusta orienting participants to stancetaking
Section 4 has shown that in the present data minun mielestä and minusta are of-
ten either embedded in or followed by an assessment, and an assessment clearly
contributes to stancetaking already on its own. An interesting question can then
be raised: What is it that makes stance markers relevant in assessments? Gener-
ally put, they orient participants to the stancetaking. More specifically, the stance
markers can be seen to perform three different functions: projecting disagreement
in a second assessment, marking transition to a first assessment across turns, and
marking transition to a first assessment within an extended turn.
5.1 Projecting disagreement in a second assessment
The few remarks that have been made on minun mielestä and minusta in previous
research mainly deal with cases in which these stance markers project different
types of disagreement (Aalto 1997: 65; Hakulinen et al. 1989: 136). Among the
assessments in the present data, however, this function is relatively rare. I present
it here first because of its clear trajectory: a speaker makes a first assessment, and
a recipient makes a disagreeing second assessment that includes a stance marker.
In example (5), Kaarina and Reija are discussing the characteristics of people
coming from different parts of Finland. Kaarina reports on what someone has
said about people who come from Savo, the area in eastern Finland that Reija is
from.

1 K: Sit se v- sano vaan et,
then it say:pst:3sg just that
‘Then he j- just said that,’
2 ...(0.3)
3 et niinku,
that prt
‘that like,’
4 (H) et ne on helv[eti-n <F>ka</F>]teellis-i-i ihmis-i-ä.
that they be:3sg hell-gen jealous-pl-ptv person-pl-ptv
‘(H) that they are damn jealous people.’
5 R: [(SNIFF)]
6 ...(0.4)
7 <HI>Mu-n miele-st se</HI> on paska-puhet-[2ta2].
I-gen mind-ela it be:3sg shit-talk-ptv
‘In my opinion it is bullshit.’
8 K: [2Se2] sano et se on niinku (H) --
it say:pst:3sg that it be:3sg prt
‘He said that it is like (H) --’
9 E- ilmeisesti se nii[3nku3],
apparently it prt
‘E- apparently it like,’
10 R: [3(SNIFF)3]
11 K: .. jotenki tule-e esiin tollase-s pikku-kaupungi-s.
somehow come-3sg into.view like.that-ine small-town-ine
‘.. somehow comes up in a small town like that.’
12 ...(0.8)
13 R: Nii joo.
prt prt
‘Oh yeah.’5
Embedded in Kaarina’s report is an assessment made by a third party: it is intro-

duced with a reporting clause on line 1, sit se v- sano vaan et ‘then he j- just said
that,’ and realized as a that-clause on line 4, et ne on helvetin kateellisii ihmisiä
‘that they are damn jealous people.’ The report does not convey the speaker’s own
stance toward the matter at hand. Nonetheless, the report is stance-inducing in
230 Mirka Rauniomaa
that it makes relevant an uptake, i.e., a display of the telling’s impact, by the recipi-
ent (see Heritage 1984). The first syllable of kateellisii ‘jealous’ is produced with
relatively loud voice, which highlights the point of the utterance and the relevance
of an uptake. A response from Reija is particularly relevant here as she represents
the group of people that has been assessed.
Kaarina’s turn accomplishes two actions that Reija can respond to, i.e., the
report as such and the assessment within it. On line 7, Reija deals with both
by producing an assessment of the reported assessment, mun mielest se on pas-
kapuhetta ‘in my opinion it is bullshit.’ The predicate noun paskapuhetta ‘shit
talk’ or ‘bullshit’ marks negative evaluation of and strong disagreement with the
third-party assessment, which is here referred to with the pronoun se ‘it.’ There
are several features in Reija’s turn that project disagreement. Firstly, Reija does
not respond immediately after Kaarina’s turn nor in overlap with it; instead there
is a 0.4-second pause on line 6, which may be interpreted as a means of stall-
ing disagreement (Pomerantz 1984: 65). Secondly, in Finnish, first assessments
are usually responded to by particles, or particle and finite verb combinations,
rather than fully-fledged second assessments (Tainio 1993: 153–154; as reported
in Tainio 1996: 108). This suggests that as soon as the utterance is recognized to
consist of something other than such a particle, i.e., in Reija’s producing mun
mielest, it can be understood to be an atypical and possibly dispreferred response
(see Pomerantz 1984 for a discussion of preference in assessments). Thirdly, the
stance marker mun mielest projects contrast by explicitly marking the start of the
speaker’s viewpoint, which in this case would not be necessary were the assess-
ment displaying agreement (see Aalto 1997: 65; Hakulinen et al. 1989: 136).
The stance marker does special work in projecting disagreement: it shows the
speaker’s sensitivity toward the fact that, on the one hand, a stance has already
been reported on, and, on the other hand, another stance is yet to be taken. In
other words, the stance marker acknowledges the existence of several possible
stances and locates the speaker’s stance among them. The fact that Kaarina from
line 8 onward defends the reported speaker by giving an account for the third-par-
ty assessment suggests that she may sympathize with both Reija’s and the reported
speaker’s stances. Had Reija not used the stance marker on line 7, the assessment
would have sounded uncompromising and left Kaarina with little opportunity to
continue: especially because Reija represents the group of people that has been
evaluated, a plain assessment by her would have borne a sense of authority and
finality, making it hard for Kaarina to differ on the issue. Starting the disagree-
ing second assessment with a stance marker, Reija leaves Kaarina more room to
manoeuvre between the reported stance and the one that Reija takes. Here, minun
mielestä indeed has a softening or modifying effect, as reported in earlier remarks
on the stance markers (e.g., Aalto 1997: 65; Hakulinen et al. 1989: 136).
The amount of disagreements that are framed with the stance markers is lim-
ited in these data. Their status is further decreased by the fact that in the case
presented as example (5) a second assessment is produced to disagree with a re-
ported speaker. However, there are also cases in which the assessment sequence
is more straightforward: in example (6), the stance marker mun mielestä proj-
ects disagreement with a co-participant. Here, Kaarina and Reija talk about Joni,
with whom they are both acquainted. Kaarina has introduced Joni into their talk
a little earlier by asking Reija to confirm a piece of information about his past
whereabouts. Kaarina has then asked several questions about him, and Reija has
provided answers as well as some extra information about his social life, which
Kaarina has marked as news to her. Reija has thus proved to be the more informed
participant on this topic by the time the participants get to the segment presented
here. The main focus is on the stance marker on line 5, but I will also briefly com-
ment on the one on line 2.6 (N.B. Reija’s first intonation unit is divided into two
misaligned lines due to limitations of space.)

1 K: Se on jotenki semmonen vähän ällö,
it be:3sg somehow such a.little disgusting
‘He is somehow a bit disgusting,’
2 <%>mu-n mie[le-stä</%>].
I-gen mind-ela
‘in my opinion.’
3 [((KNOCK))] ((KNOCK))
4 R: (H) Mut se-ki on %m= --
but it-cli be:3sg
‘(H) But it/he is also m= --’
5 .. %Se on mu-n miele-stä ol-lu pitkään
it be:3sg I-gen mind-ela be-pcp long
‘.. In my opinion he has been damn disgusting
helveti-n ällö mut,
hell-gen disgusting but
for a long time but,’
6 ((RATTLE))
7 R: oikeestaan ku sii-hen tutustu-u paremmin,
actually when it-ill acquaint-3sg better
‘actually when Ø gets to know him better,’
232 Mirka Rauniomaa
8 ni se ällöys vähene-e.
prt it disgustingness diminish-3sg
‘the disgustingness diminishes.’
On line 1, Kaarina makes an assessment of Joni: se on jotenki semmonen vähän

ällö ‘he is somehow a bit disgusting.’ There is some material, i.e., the planning par-
ticle jotenki ‘somehow’ (see ISK 2004: 821–822) and the demonstrative adjective
semmonen ‘such’ (see Erringer 1996; as cited in Helasvuo 2004: 1327), which seem
to signal hesitation and to give Kaarina time to plan her speech. On line 2, Kaa-
rina then adds the stance marker mun mielestä as a separate intonation unit and
in softer speech. This is one of the two instances in the data in which the stance
marker minun mielestä is added after the proposition as an intonation unit of its
own. Taking into consideration that Reija has a moment earlier been identified
as the more informed participant on the topic, it seems evident that Kaarina uses
mun mielestä to mark her assessment as a possibly disputable one. The first as-
sessment already in itself invites a subsequent contribution, but the added stance
marker puts even more emphasis on the relevance of a response from a co-partici-
pant who clearly has better access to the assessable.
On line 4, Reija starts an utterance with the particle mut ‘but,’ which projects
contrast (see Sorjonen 1989: 176). However, the intonation unit is truncated, and
on line 5, Reija redesigns her utterance into a second assessment that seemingly
agrees with Kaarina’s, se on mun mielestä ollu pitkään helvetin ällö mut ‘in my
opinion he has been damn disgusting for a long time but.’ Kaarina’s and Reija’s
turns can be mapped onto a diagraph, which is used “to create a representation
of the structure of resonance relations that arise between utterances” (Du Bois
2003a). Diagraph 1 shows how Reija recycles elements from Kaarina’s turn.
Diagraph 1. Recycled and upgraded elements in (6)

1 K: Se on jotenkin semmonen vähän ällö,
2 mun mielestä.
5 R: Se on mun mielestä ollu pitkään helvetin ällö mut,
Reija’s assessment is an upgrade of Kaarina’s assessment in three senses: Firstly,

Reija adds a timespan into Joni’s disgustingness by using the present perfect form
on ollu ‘has been’ and the adverbial pitkään ‘for a long time.’ Secondly, unlike in
Kaarina’s turn, there are no signs of hesitation in Reija’s turn. Thirdly, Reija inten-
sifies the evaluative adjective ällö ‘disgusting’ with the amplified adverbial helvetin
‘damn’ rather than the toned-down vähän ‘a bit.’ The stance marker mun mielestä
in Reija’s utterance on line 5 attributes the upgrades to the current speaker and
highlights the emerging stance differential. Reija’s mun mielestä responds directly
to Kaarina’s use of the stance marker by heightening the resonance between the
participants’ turns and consequently draws attention to the differences in their
turns. It signals that Reija is not going to take up the position that is already avail-
able but will evoke another one.
The contrast that was implied by mut ‘but’ at the start of line 4 is reinstated
at the end of line 5. In this light, Reija’s assessment can be seen as a report of
her once, but no longer, having taken a similar stance as Kaarina takes now. The
use of the present perfect on ollu ‘has been’ can now be interpreted as projecting
Reija’s changed state (see Sacks 1992: 181–182, who calls certain forward-project-
ing verbs, such as thought and wanted to, ‘first verbs’). Reija goes on to specify the
new state and argues on lines 7 and 8 that once better acquainted with Joni, he
does not seem so disgusting anymore. It is worth noting that Reija uses the zero-
person construction on line 7, oikeestaan ku siihen tutustuu paremmin ‘actually
when Ø gets to know him better,’ to imply that anyone, including Kaarina, may in
the long run change their view of Joni (see Laitinen 1995 for a discussion of the
zero-person construction in Finnish).
In sum, the two minun mielestä in example (6) orient the participants to the
stancetaking: the first one by anticipating alignment and the second one by re-
sponding to that anticipation. As in example (5), the second stance marker in ex-
ample (6) shows the speakers’ sensitivity toward the existence of several possible
stances. Such framing does not seem to be necessary when participants’ stances
converge: indeed, the present data do not include any agreeing second assess-
ments that would be framed with minun mielestä or minusta. Here, the stance
markers are placed at the start of a disagreeing second assessment, i.e., they subtly
project disagreement in a context in which an explicit evaluation has already been
made. More often, however, the stance markers are used at points of transition
from non-evaluative or implicitly evaluative to explicitly evaluative talk. Such cas-
es form the majority of the data and can be divided into two equally-sized groups:
one in which the transition takes place across speakers’ turns, as in the next set of
examples, and one in which there is a transition within a speaker’s extended turn,
as in Section 5.3.
5.2 Marking transition to a first assessment across speakers’ turns
The stance markers minun mielestä and minusta are also found in contexts in
which a speaker makes an assessment about a topic that has not been evaluated
or has only been implicitly evaluated in the talk so far. When the shift takes place
across turns (i.e., between speakers), the stance markers can be used to signal the
234 Mirka Rauniomaa
start of a new action and to prepare the participants for relevant next actions that
further contribute to the stancetaking.
In example (7), Ira_1 explicates to her co-participants how one can boil water
and breathe in the warm steam in order to clear one’s respiratory system when
suffering from a cold.
(7) sg151_b2 Chat about Something <T:00:08:47>

1 I1: Mut si-tä pitää niin kauan (H),
but it-acc have.to so long
‘But Ø has to keep on (H),’
2 ...(0.3)
3 pitä-ä tai hengittä-ä ja,
hold-inf or breathe-inf and
‘holding or breathing it and,’
4 aina keittä-ä k_uut-ta vet-tä et,
always boil-inf new-acc water-acc that
‘always boil fresh water until,’
5 tunte-e et tää-lt niinku,
feel-3sg that here-abl prt
‘Ø feels that like out of here,’
6 ...(0.3)
7 [rupe-e irto-o-n,
start-3sg come.off-inf-ill
‘Ø starts coming off,’
8 E: [lähte-e <X>irki</X>],
come-3sg loose
‘Ø comes loose,’
9 I1: sie-lt] tule-e semmos-i-i kökkö-j-ä.
there-abl come-3sg such-pl-ptv lump-pl-ptv
‘these lumps come out of there.’
10 ...(0.8)
11 I2: (H) Mu-st se on hirvee-n inhottava ol-la jossain (H),
I-ela it be:3sg awful-gen nasty be-inf somewhere
‘(H) I think it is awfully nasty to be in some (H),’
12 ä- <A>semmose-n</A> kuuma-n katti[la-n X,
such-gen hot-gen kettle-gen
‘ä- X such a hot kettle,’
13 I1: [Mm].
14 I2: al]ka-a hiki va[2lu-a ja2] [3kai3]kke-e.
start-3sg sweat run-inf and everything-ptv
‘sweat starts running and everything.’
15 I1: [2Mm2].
16 E: [3Joo3].
prt
‘Yeah.’
The topic of steam breathing has started with Ira_1’s giving advice to Emma, who
is evidently suffering from a cold, and Emma’s alignment as the advice recipient.
On lines 1–7 and 9, Ira_1 gives a more detailed description of how the advice
should be carried out. The use of the zero-person construction on lines 1–5 makes
the actor’s role available to Emma without explicitly attributing it to her (see
Laitinen 1995): mut sitä pitää niin kauan pitää tai hengittää ja aina keittää k_uutta
vettä et tuntee et täält niinku ‘but Ø has to keep on holding or breathing it and always
boil fresh water until Ø feels that like out of here.’ When Ira_1 pauses to search for a
word on line 6, Emma comes in on line 8 with a syntactically fitted continuation
of Ira_1’s utterance, lähtee <X>irki</X> ‘Ø comes loose’ (see Helasvuo 2004 for a
discussion of co-constructions in Finnish). Emma’s coming in at this point marks
a successful receipt of the advice and implies readiness to end the sequence. In-
deed, once Ira_1 has finished her overlapping utterance on line 9, a 0.8-second
pause follows on line 10.
On lines 11–12, Ira_2 makes an assessment that is on the topic but provides
a slightly different aspect to it: must se on hirveen inhottava olla jossain ä- sem-
mosen kuuman kattilan X ‘I think it is awfully nasty to be in some ä- X such a hot
kettle.’ As the participants’ focus has so far been on the positive effects of steam
breathing, it is potentially precarious to evaluate it negatively. The stance marker
must at the start of Ira_2’s utterance on line 11 addresses the issue: it frames the
assessment as being made strictly from one particular point of view and thus pre-
empts potential disagreement. It also provides Ira_2 a way in to participating in
topical talk, although she has not been the main recipient of her co-participants’
previous turns and the sequence was coming to an end. This interpretation is
reinforced by the fact that Ira_1’s minimal response tokens on lines 13 and 15, as
well as Emma’s quiet response token joo on line 16 do not mark strong alignment
with, but simply acceptance of, Ira_2’s stance (see Sorjonen 2001: 204–205 for a
discussion of joo).
In example (8) the same participants are discussing the benefits of going to
a tanning salon before leaving for holidays to a sunnier and warmer part of the
236 Mirka Rauniomaa
world. Emma has been particularly active in reporting on an article that she has
read on the issue and in arguing for the need to use the sun bed several times for
it to have any significant effect.

1 I1: Mut silleen et,
but so that
‘But so that,’
2 .. et nyt,
that now
‘.. that now,’
3 .. pari kolme kerta-a käy-ä-än ennen [ku lähe-tä-än].
couple three time-ptv go-pass-4 before than leave-pass-4
‘.. we go two or three times before we leave.’
4 E: [<X><HI>Nii joo</HI></X>],
prt prt
‘Oh yeah,’
5 ni se on ihan --
so it be:3sg quite
‘so it is quite --’
6 Miu-st se on [kans ihan järke]vä-ä,
I-ela it be:3sg also quite sensible-ptv
‘I also think it is quite sensible,’
7 I1: [<X>ettei ihan pala</X> (Hx)].
that-neg quite burn
‘so that Ø doesn’t burn completely.’
A moment earlier, Ira_1 has been told that she still has a tan left from her previ-
ous trip; on lines 1–3, she goes on to state that she and her travel partner will
nevertheless go to a tanning salon before their next trip (in order to get a more
protective tan). In overlap with the end of Ira_1’s utterance, Emma displays a kind
of realization with the high-pitched nii joo ‘oh yeah.’ The realization is most likely
induced by the fact that whereas Emma has been arguing for the benefits of the
sun bed when one has not had exposure to the sun for a while, Ira_1 has here im-
plied that going to a tanning salon may prove useful even when one still has an old
tan. In any case, Emma uses the opportunity to give more weight to the point that
she has been making about the necessity of using the sun bed frequently enough.
On line 6, Emma assesses Ira_1’s proposed line of action in view of what she has
said earlier: miust se on kans ihan järkevää ‘I also think it is quite sensible.’
It is worth noticing that Emma starts to make her assessment already on line
5 but cuts it off before producing the assessment term, ni se on ihan ‘so it is quite.’
She repairs the assessment by adding the stance marker miust and the alignment
marker kans (the Finnish kans is similar to the English too, which Du Bois 2003b
calls an ‘intersubjective alignment marker’). The completed assessment on line
6 is more in line with the general progress of the sequence: miust attributes the
assessment to the current speaker, while kans recognizes an implicit evaluation
in the previous speaker’s turn (because people usually engage themselves only in
what they consider sensible).
In examples (7) and (8), the stance markers signal a transition from earlier,
non-evaluative, or implicitly evaluative talk by one speaker to an explicit evalua-
tion by another. Speakers manage such transitions without using minun mielestä
or minusta, but employing a stance marker helps them to display an understand-
ing of the sequential context in which they produce the assessment. The stance
markers subtly deal with the risk of disagreement, reprimand, etc. that partici-
pants take in moving from non-evaluative or implicitly evaluative to explicitly
evaluative talk across turns. In the following section, the focus will be on transi-
tions within speakers’ own talk.
5.3 Marking the transition to a first assessment within an extended turn
The third function that the stance markers have in the present data is to mark a
transition from non-evaluative or implicitly evaluative talk to an explicit evalua-
tion by the same speaker. There are two cases in the data in which minun mielestä
and minusta mark the transition to a parenthetical assessment that requires only
minimal uptake by a recipient (see Duvallon and Routarinne 2001 for a discus-
sion of parenthesis in Finnish). More typically, however, shifts from non-evalu-
ative or implicitly evaluative to explicitly evaluative talk occur in extended turns
when a speaker moves from some kind of a telling that has received only minimal
response tokens from the recipients, to an evaluative summary that invites more
substantial contributions from the co-participants. In the following two examples,
an uptake is clearly expected after the assessment.
Reija and Kaarina are both lawyers and talk about work-related issues several
times during the conversation. Here, Kaarina reports on how some of her col-
leagues treat witnesses in court.
238 Mirka Rauniomaa

1 K: Mut mei-l on niinku hirvee-t <X>kähmä-ä</X> sii-tä ku,
but we-ade be:3sg like awful-ptv fight-ptv it-ela that
‘But we have like awful fights over the fact that,’
2 (TSK) (H) koska mä e-n ymmärrä si-tä että,
because I neg-1sg understand it-acc that
‘(TSK) (H) because I don’t understand it that,’
3 .. että [.. pyyde-tä-än] todistaja sisään,
that ask-pass-4 witness in
‘.. that .. (they) ask the witness in,’
4 [((BANG)) ((BANG))]
5 K: ja sit lue-ta-an si-lle esitutkintapöytä[2kirja-a2],
and then read-pass-4 it-all record.of.preliminary.investigation-acc
‘and then (they) read him/her the record of the preliminary investigation,’
6 R: [2(SNIFF)2]
7 K: ja kysy-tä-än [3että u3]udista-tte-ko [4tämä-n4].
and ask-pass-4 that renew-2pl-q this-acc
‘and ask (him/her) that do you confirm this.’
8 R: [3(SNIFF)3]
9 [4Nii4].
prt
‘Yeah.’
10 K: .. Se on mu-n miele-st ihan per[5see-stä.
it be:3sg I-gen mind-ela quite ass-ela
‘.. In my opinion it is really bollocks.’
11 R: [5Väärin.
wrong
‘Wrong.’
12 K: Tommonen asia5].
like.that matter
‘A thing like that.’
13 R: Niin-hän se on-ki5].
so-cli it be:3sg-cli
‘So it is.’
Kaarina starts a complaint sequence on line 1 by stating that she and her col-
leagues have grossly divergent views on a court procedure. On line 2, she specifies
her own take on the matter, i.e., not understanding, and then goes on to describe
the contested procedure on lines 3, 5, and 7. Kaarina can be seen to seek support
from Reija, who is able to view the matter from the same professional perspective
as Kaarina. Reija’s partly overlapping response token nii on line 9 implies that
Kaarina has secured her alignment: “nii agrees with the prior talk by claiming
recognition of what the co-participant is talking about” (Sorjonen 2001: 247).
Kaarina ends her complaint on line 10 with an “overt expression of moral
indignation” (Drew 1998: 309–311) by making an assessment that wraps up her
complaint: se on mun mielest ihan perseestä ‘in my opinion it is really bollocks.’
The stance marker mun mielest is positioned in a transition point between im-
plicit and explicit evaluation. It helps Kaarina to construct her own stance in rela-
tion to and in contrast with all the possible ones that she has referred to a moment
earlier. It is worth noting that in English the expression of indignation at the end
of a complaint sequence usually takes the form of a first-person assessment, such
as I was so angry (Drew 1998: 311); here, the assessment is personalized with the
stance marker mun mielest.
On line 11, Reija displays recognition of the import of Kaarina’s utterance and
reciprocates an expression of indignation. As is typically the case with agreements
to assessments, Reija produces her contribution in slight overlap with Kaarina’s
utterance (see Pomerantz 1984: 69). More specifically, following the common
practice in Finnish, the overlap starts on the assessment term perseestä ‘from the
ass’ or ‘bollocks’ rather than on the adverbial intensifier ihan ‘really’ (see Tainio
1996: 89; cf. Goodwin and Goodwin 1987: 22–24 for American English). Kaarina
has by this time said enough for Reija to be able to produce an overlapping assess-
ment term that syntactically fits in with Kaarina’s ongoing utterance.
Similar to example (9), in example (10), the stance marker projects an ex-
plicit evaluation that invites further contributions, but the co-participants’ uptake
is somewhat delayed. The example starts with Ira_1’s pointing out to Ira_2 and
Emma that she and some others have reserved a sauna.

1 I1: <HI>Ties-i-k-s Ira</HI> muuten et me vara-tt-i-in se
know-pst-q-cli first_name by.the.way that we reserve-pass-pst-4 it
‘By the way Ira did you know that we reserved the
.. Lappee-n …(0.4) sauna.
place_name-gen sauna
.. sauna … at Lappee.’
240 Mirka Rauniomaa
2 I2: Joo.
prt
‘Yeah.’
3 Kuul-i-n Tu[pu-lta].
hear-pst-1sg first_name-abl
‘I heard from Tupu.’
4 I1: [Tapanin]päivä-ks.
Boxing.Day-trans
‘For Boxing Day.’
5 I2: J[2oo2].
prt
‘Yeah.’
6 E: [2Jo2]o.
prt
‘Yeah.’
7 ((CLANK))[=]
8 I1: [Viis kym]ppi-i maksa-a kolme tunti-a,
five ten-ptv cost-3sg three hour-ptv
‘Three hours cost fifty marks,’
9 ei miu-st se oo hirvee-n paha.
neg I-ela it be awful-gen bad
‘I don’t think it is awfully high.’
10 ...(0.4)
11 [Koska otta-a huomio-on] että,
because take-3sg account-ill that
‘Because if Ø takes into account that,’
12 I2: [Ei oo,
neg be
‘No it isn’t,’
13 <X>monelta</X>].
at.what.time
‘at what time.’
On lines 1–6, Ira_1 repeatedly provides her co-participants with pieces of infor-
mation that might be news to them, but that they mark as known information.
On line 8, Ira_1 provides yet another piece of information, i.e., the cost of renting
the sauna. This time, however, Ira_1 does not wait for any response from Ira_2
or Emma but continues on line 9 to assess the piece of information she has just
given, ei miust se oo hirveen paha ‘I don’t think it is awfully high.’ So once it has
become clear that the participants share the information about the sauna, Ira_1
moves on to a logical next action and uses an explicit means to do so. The stance
marker miust signals the transition and the upcoming relevance of further con-
tributions. It implies that Ira_1 makes the assessment specifically in the role that
she is holding here, i.e., as the one who has reserved the sauna and provided the
others information about its price.
Ira_1’s assessment is followed by a 0.4-second pause on line 10, which would
allow for an uptake by Ira_2 or Emma. As no such uptake ensues, Ira_1 starts on
line 11 to account for the assessment that she made on line 9. However, at the very
same moment Ira_2 comes in with a second assessment on line 12, ei oo ‘no it isn’t.’
Consisting simply of the negated predicate verb that Ira_1 used in her assessment,
Ira_2’s turn implies strong agreement with Ira_1’s (see Tainio 1996: 97–99), but
it is somewhat obscured by the overlap. After the segment presented here, the
participants continue to display their agreement although overlapping remains a
potential problem for a while.
In sum, the stance marker miust is here placed at a point in which it marks
one participant’s explicit shift to stancetaking and signals an upcoming opportu-
nity for the other participants to contribute. The co-participants display recogni-
tion of these functions in their belated responses. Examples (9) and (10) have
shown that the stance markers can be used at transition points from non-evalu-
ation or implicit evaluation to an explicit evaluation of a state of affairs. It is true
that participants can make such shifts without using stance markers, but by us-
ing them speakers can orient their recipients to what follows, especially if the
speaker’s ongoing action will require recipients’ response. I will discuss this issue
in more detail in the following section.
6. Evidence from a special case: Stance markers reflecting

interactional contingencies
Previous examples have shown that speakers use the stance markers minun
mielestä and minusta to orient their co-participants to the stancetaking. The
stance markers are made relevant by a need to mark stancetaking explicitly at
some interactional moment. Example (11) will provide extra evidence: more or
less the same assessment is made first without and then with a stance marker (see
Schegloff 1996: 192–199, who encourages analysts to pay attention to ‘eventful
nonoccurrence’).
Before the selected segment, Kaarina and Reija have compared two possible
ways of putting forward payment orders. Kaarina then moves on to a slightly
242 Mirka Rauniomaa
different topic by showing her appreciation of the fact that Reija will soon start
working at the same workplace with her.

1 K: (TSK) (H) Se on helveti-n hyvä ku sä [tuu-t sinne],
it be:3sg hell-gen good when you come-2sg there
‘(TSK) (H) It is damn good that you will come there,’
2 R: [(Hx)]
3 K: ku su-lla on kuitenki se-n verran hallu-s
because you-ade be:3sg anyway it-gen much possession-ine
‘because after all you master those things’
noi jutu-t ni,
those thing-pl prt
‘to such an extent,’
4 (H) kato ku se on si-llä [tava-l]la niin että,
look:imp:2sg because it be:3sg it-ade way-ade so that
‘(H) see because it is so that,’
5 R: [(SNIFF)]
6 K: .. ne ei niinku --
they neg prt
‘.. they don’t like --’
7 ...(0.8)
8 Nää nuore-mma-t notaari-t,
these young-comp-pl notary-pl
‘These younger notaries,’
9 ni ne ei,
prt they neg
‘they don’t,’
10 tiä-t-kö se teke-e se-n esitysehdotukse-nsa periaat[tee-ssa]
know-2sg-q it make-3sg it-acc proposal-acc:poss principal-ine
‘you know he/she makes his/her proposal basically’
vaan sii-hen,
just it-ill
‘just there,’
11 R: [Joo],
prt
‘Yeah,’
12 K: .. Kirjotta-a vaan sii-hen niinku,

write-3sg just it-ill prt
‘.. Just writes it there like,’
13 .. sii-hen hakemukse-n pääl- päälle,
it-ill application-gen on.top
‘.. on top- on top of the application,’
14 että miten se niinku kannatta-i[s <X>an]-taa</X> ni,
that how it prt be.beneficial-cond give-inf prt
‘that how Ø should like give it,’
15 R: [Nii],
prt
‘Yeah,’
((17 lines omitted, re: Kaarina’s colleagues))
On line 1, Kaarina makes an assessment, se on helvetin hyvä ku sä tuut sinne ‘it

is damn good that you will come there.’ Perhaps partly because of the constraints
on self-praise (see Pomerantz 1978), Reija responds to Kaarina’s positive assess-
ment only with a somewhat forceful exhalation on line 2. There are two features in
Kaarina’s turn that indicate that no elaborate uptake is expected: Kaarina’s utter-
ance on line 1 ends in a fairly level intonation that implies continuity, and there is
no noticeable pause before her next utterance on line 3. Indeed, immediately after
making the assessment, Kaarina continues on line 3 to account for the assessment
and to praise Reija even further: ku sulla on kuitenki sen verran hallus noi jutut ni
‘because after all you master those things to such an extent.’ The assessment works
as an introduction to the current topic, which Kaarina begins to elaborate.
From line 4 onwards, Kaarina provides an extended account that contrasts
Reija’s competence with the incompetence of Kaarina’s current colleagues who are
somewhat negligent in making their proposals so that others will have to check
their work afterwards. Reija occasionally displays acknowledgement or agree-
ment, while Kaarina continues her lengthy account. There is then a moment of
heavy overlap before Kaarina comes back to her previous point:
(11)
16 K: (TSK) (H) Ni --
prt
‘(TSK) (H) So --’
17 ...(0.5)
244 Mirka Rauniomaa
18 <X>Tosiaan m-</X> --
really
‘Really i- --’
19 <A>Mu-n miele-stä</A> on niinku tosi hyvä et sä tuu-t.
I-gen mind-ela be:3sg prt really good that you come-2sg
‘In my opinion it is like really good that you will come.’
20 Koska,
because
‘Because,’
21 .. e-n mä rupee niinku nii-t silleen nyhrä-ä.
neg-1sg I start prt they-acc that.way tinker-inf:ill
‘.. I won’t start tinkering with them.’
22 Mu-lla men-is sii-hen niinku,
I-ade go-cond:3sg it-ill prt
‘It would take me like,’
23 <X>mä nää-n se-n Heiki-n,
I see-1sg it-acc first_name-acc
‘I see Heikki,’
24 si-l on</X> niin älyttömä-sti tö-i-tä,
it-ade be:3sg so unreasonable-adv work-pl-ptv
‘he has such an unreasonable amount of work to do,’
25 (H)[= et] niinku,
that prt
‘(H) that like,’
26 R: [Nii],
prt
‘Yeah,’
27 K: mä e-n rupee [2tollas2]-t-en kanssa [3nyh3]rää-[4mä-än4].
I neg-1sg start that.kind-pl-gen with tinker-inf-ill
‘I won’t start tinkering with those.’
28 R: [2(SNIFF)2] [3(SNIFF)3]
29 [4Eli su-l4]la pitää ol-la joku jo-hon
so you-ade have.to be-inf someone who-ill
‘So you have to have someone
sä voi-t [5luot-taa5].
you can-2sg trust-inf
you can trust.’
30 K: [5Nii5].
prt
‘Yeah.’
31 Ni[6menomaan6].
exactly
‘Exactly.’
32 R: [6(TSK) (H) Nii6].
prt
‘(TSK) (H) Yeah.’
The overlap, which has been omitted here, may be one reason for the trouble that
Kaarina has in resuming talk on lines 16–18. Although both intonation units are
truncated, the particles ni ‘so’ and tosiaan ‘really’ succeed in projecting that what
Kaarina will say next is tied to what she has said before (see Hakulinen 1997: 45
for a remark on ni). The connection becomes evident on line 19, when Kaarina
makes a similar assessment as she did on line 1, mun mielestä on niinku tosi hyvä
et sä tuut ‘in my opinion it is like really good that you will come.’ Diagraph 2
shows the lexical and syntactic similarities and differences between the two as-
sessments.
Diagraph 2. Speaker’s modified repetition of an assessment in (11)

1 K: Se on helvetin hyvä ku sä tuut sinne,
19 K: mun mielestä on niinku tosi hyvä et sä tuut.
The same assessment term, hyvä ‘good,’ is used in both assessments, but the ad-
verbial intensifier helvetin ‘damn’ on line 1 is replaced with the equally strong tosi
‘really’ on line 19.6 The actual subject is realized as a complement clause in both
assessments, but in the second one the optional dummy subject se ‘it’ is omitted in
front of the copula and the conjunction ku ‘when’ is replaced with et ‘that.’ How-
ever, these changes do not affect the import of line 19 in comparison with line 1,
and neither does the adverbial sinne ‘there,’ which is left out from line 19 possibly
because the referred place has already become well-established, nor the particle
niinku ‘like,’ which is added to the second assessment to give Kaarina time to plan
her speech (ISK 2004 includes niinku in the group of ‘planning particles’).
The most interesting difference between Kaarina’s assessments on lines 1 and
19 is the inclusion of the stance marker mun mielestä, which is placed at the start
of the intonation unit. The fact that Kaarina repeats the assessment confirms her
stance, but at the same time the addition of mun mielestä implies a shift in the
stancetaking. When Kaarina from line 20 onward accounts for the assessment
246 Mirka Rauniomaa
that she has just made, she brings in a more personal aspect: rather than compar-
ing Reija with her colleagues as she did from line 3 onward, Kaarina now expli-
cates how she would benefit from having Reija around.
Once Kaarina has brought in the personal aspect, Reija can contribute to the
stancetaking more freely, as her subsequent turn indicates. On line 29, Reija name-
ly produces a formulation (Heritage and Watson 1979), that is clearly marked as
such with the particle eli ‘so’: eli sulla pitää olla joku johon sä voit luottaa ‘so you
have to have someone you can trust.’ In this way, without making direct reference
to herself, Reija displays agreement with the positive assessment that Kaarina has
made of her. On lines 30–31, Kaarina accepts Reija’s formulation and consequent-
ly reinforces her own stance.
In moving from an unmarked assessment to one that includes the stance
marker minun mielestä, the participants of example (11) engage themselves quite
differently in the stancetaking. A somewhat restricted kind of stancetaking is
gradually opened up, and a one-sided praise turns into mutual positive evalu-
ation. The stance marker mun mielestä works here to orient the participants to
this change. To put it more generally, minun mielestä and minusta reflect the con-
tingencies that arise in interaction: framing an assessment with minun mielestä
or minusta provides speakers a means of taking into consideration the sequen-
tial environment in which they produce their utterance as well as any emerging
changes in it.
7. Conclusion
The purpose of this paper was to contribute to the understanding of the Finnish
stance markers minun mielestä and minusta. In Section 4, I first showed that there
is significant regularity in the occurrence of the stance markers. At a more local
level, they are placed either at the start or, to a lesser extent, in the middle of the
intonation unit. The range of elements that can be placed before the stance marker
in the intonation unit is very limited, e.g., in assessments, the evaluative element
never precedes the stance marker. At a sequential level, the stance markers can be
placed in either initiating or responding positions. This variety of sequential posi-
tions is made possible by the types of action that the stance markers frequently
occur in, e.g., assessments.
Rather than attempting at a comprehensive overview of minun mielestä and
minusta across different action types, I then focused on the stance markers as they
occur in the context of assessments. I argued that as stancetaking is not limited to
such explicit means as evaluative, lexical and grammatical items or the evaluative
action of assessing, participants’ use of a stance marker within an assessment is

brought about by an interactional requirement.
I identified three orientational functions for minun mielestä and minusta:
projecting disagreement in a second assessment, marking transition to a first
assessment across speakers’ turns, and marking transition to a first assessment
within an extended turn. In such environments, the stance markers can be seen
to deal with the potential riskiness of disagreeing or moving from non-evaluative/
implicitly evaluative to explicitly evaluative talk. They do so by acknowledging the
existence of several possible stances and showing recognition of the sequential
context in which the assessment is produced. That is to say, they display an in-
tersubjective understanding in a demonstrably subjective environment and thus
facilitate recipients’ stance taking.
Stancetaking can very well be managed without any explicit linguistic fram-
ing, but the fact that speakers make similar assessments with and without stance
markers argues for an appreciation of minun mielestä and minusta as an interac-
tionally motivated means of orienting participants to stancetaking. What is more,
studying specific linguistic items as they occur in actual talk-in-interaction con-
tributes to an understanding of stancetaking as the participants’ joint endeavour.
Notes
1. I wish to thank John Du Bois, Robert Englebretson, Pentti Haddington, Tiina Keisanen,
Elise Kärkkäinen, Maarit Niemelä, Liisa Raevaara, Sanja Starck, as well as the participants of
the 10th Biennial Rice Linguistics Symposium for the inspiring and encouraging atmosphere
in which I was able to prepare this paper. I am also grateful to the two anonymous reviewers
for examining my paper and suggesting improvements. This study was made possible by fund-
ing from the Academy of Finland (project number 53671) and Langnet, the Finnish Graduate
School in Language Studies.
2. To be exact, the fullest form of the expression is minun mielestäni, in which both the per-
sonal pronoun minun and the first-person-singular possessive suffix -ni are used. In the present
data, and in spoken Finnish more generally (Ikola 1992, as cited in Paunonen 1995; Paunonen
1995), the possessive suffix is not used at all. However, see Luukka (1992a: 119) for a discus-
sion of the varied forms, including mielestäni, that are used in spoken and written academic
discourse.
3. However, there are two cases of minun mielestä in which most items precede the stance
marker, and only the object of the transitive clause that comprises the intonation unit follows
the stance marker – examples (a) and (b).
(a) sg067_a1 Lawyer Issues <T:00:03:46>

R: Se anta-a hirmu hyvä-n kuvan mun miele-st sii-tä leffa-sta,
it give-3sg really good-acc picture-acc I-gen mind-ela it-ela film-ela
‘It gives a really good idea in my opinion of the film,’
248 Mirka Rauniomaa
(b) sg151_a Chat about Something <T:00:40:07>

I1: Mut si-llä-hän voi käy-dä ostaa mi-un miele-st nii-t,
but it-ade-cli can go-inf buy:inf:ine I-gen mind-ela those-acc
‘But you can use it to buy in my opinion those,’
These two cases offer interesting evidence of how contingencies are managed in conversation:
although relatively fixed as regards their position in the intonation unit, the stance markers
offer a flexible resource for participants to address some interactionally relevant issue, which
may or may not have been material at the moment they started speaking. For a discussion of
contingency, see Ford (2004).
4. See Helasvuo (2001b: 78) for a discussion of the verb- and particle-like features of the nega-
tion element ei.
5. In this paper, I have translated the response particles nii and joo as ‘yeah.’ However, Sor-
jonen (1999, 2001) has shown that they are in effect used in different sequential contexts. As the
translation of nii joo in example (5) implies, their combination serves yet another interactional
purpose.
6. The change may partly be inspired by the use of tosiaan on line 18. I thank Elise Kärkkäinen
for this observation.
References
Aalto, E. 1997. Mielipiteen vuorovaikutuksellinen rakentaminen syntyperäisen ja ei-syntype-

räisen suomenpuhujan keskustelussa [Interactional Construction of Opinion in Conversa-
tions between Native and Non-Native Speakers of Finnish]. Licentiate’s Thesis, Department
of Finnish, University of Jyväskylä.
Bybee, J. and Hopper, P. 2001. “Introduction to frequency and the emergence of linguistic
structure.” In Frequency and the Emergence of Linguistic Structure, J. Bybee and P. Hopper
Dahl, Ö. 2000. “Egocentricity in discourse and syntax.” Functions of Language 7(1): 37–77.
Drew, P. 1998. “Complaints about transgressions and misconduct.” Research on Language and
Social Interaction 31 (3, 4): 295–325.
Du Bois, J.W. 2003a. “Dialogic syntax: The syntax of engagement.” Paper presented at Langnet
Seminar, Oulu, September 10.
Du Bois, J.W. 2003b. “The stance differential.” Paper presented at Stance Taking, Dialogic Syntax
and Interaction Seminar, Oulu, September 8.
Du Bois, J.W., Cumming, S., Schuetze-Coburn, S. and Paolino, D. (eds.). 1992. Discourse Tran-
scription. Santa Barbara Papers in Linguistics, Vol. 4. Department of Linguistics, Univer-
sity of California, Santa Barbara.
wards and M.D. Lampert (eds.), 45–89. Hillsdale, NJ: Lawrence Erlbaum Associates.
Duvallon, O. and Routarinne, S. 2001. “Parenteesi keskustelun kieliopin voimavarana [Pa-
renthesis as a grammatical resource in conversation].” In Kieli 13: Keskustelunanalyysin
näkymiä [Perspectives in Conversation Analysis], M. Halonen and S. Routarinne (eds.),
122–154. Helsinki: Department of Finnish, University of Helsinki.
Erringer, A. 1996. The Functions of Demonstrative Adjectives Semmoinen, Tämmöinen and

Tuommoinen in Finnish Conversations. Master’s Thesis, Department of Linguistics, Uni-
versity of Colorado.
Ford, C., Fox, B.A. and Thompson, S.A. (eds.). 2002. The Language of Turn and Sequence. New
Ford, C.E. 2004. “Contingency and units in interaction.” Discourse Studies 6 (1): 27–52.
Goodwin, C. and Goodwin, M.H. 1987. “Concurrent operations on talk: Notes on the interac-
tive organization of assessments.” IPrA Papers in Pragmatics 1 (1): 1–54.
Goodwin, C. and Goodwin, M.H. 1992. “Assessments and the construction of context.” In Re-
Hakulinen, A. 1997. “Vuorottelujäsennys [Turn organization].” In Keskustelunanalyysin pe-
rusteet [Introduction to Conversation Analysis], L. Tainio (ed.), 32–55. Tampere: Vas-
tapaino.
Hakulinen, A., Suojala, M. and Sorjonen, M.-L. 1989. “Partikkelit ja muut kiteymät vuoroissa
[Particles and other clusters in conversational turns].” In Kieli 4: Suomalaisen keskustelun
keinoja I [Characteristics of Finnish Conversation I], A. Hakulinen (ed.), 115–146. Depart-
ment of Finnish, University of Helsinki.
Helasvuo, M.-L. 2001a. “Emerging syntax for interaction: Noun phrases and clauses as a syn-
tactic resource for interaction.” In Studies in Interactional Linguistics, E. Couper-Kuhlen
and M. Selting (eds.), 25–50. Amsterdam: John Benjamins.
Helasvuo, M.-L. 2001b. Syntax in the Making: The Emergence of Syntactic Units in Finnish Con-
versation. Amsterdam: John Benjamins.
Helasvuo, M.-L. 2004. “Shared syntax: The grammar of co-constructions.” Journal of Pragmatics
36 (8): 1315–1336.
tures of Social Action, J.M. Atkinson and J. Heritage (eds.), 299–345. Cambridge: Cam-
bridge University Press.
Heritage, J.C. and Watson, D.R. 1979. “Formulations as conversational objects.” In Everyday
Language: Studies in Ethnomethodology, G. Psathas (ed.), 123–162. New York: Irvington
Publishers.
Ikola, O. 1992. “Omistusliitteiden harvinaistuminen [Rarefaction of possessive suffixes].” In
Yhteiskunta muuttuu – kieli muuttuu. Nykysuomen Seuran 10-vuotisjuhlakirja [Society
Changes – Language Changes. The 10th Anniversary Volume of the Association of Contem-
porary Finnish], V. Yli-Vakkuri, M. Länsimäki and A. Nyman (eds.), 176–196. Porvoo:
WSOY.
ISK 2004. Iso suomen kielioppi [A Comprehensive Grammar of Finnish], A. Hakulinen, M.
Vilkuna, R. Korhonen, V. Koivisto, T.R. Heinonen and I. Alho (eds.). Helsinki: Finnish
Literature Society.
Karlsson, S. 2003. “Interactional uses for an epistemic marker: The case of ‘jag tycker’/’tycker
jag’ in Swedish.” Melbourne Papers in Linguistics and Applied Linguistics 3 (1): 5–23.
Keevallik, L. 2003. From Interaction to Grammar: Estonian Finite Verb Forms in Conversation.
Ph.D. dissertation. Department of Finno-Ugric Languages, Uppsala.
Kärkkäinen, E. 2006. “Stance-taking in conversation: From subjectivity to intersubjectivity.”
Text and Talk 26 (6): 699–731.
250 Mirka Rauniomaa
Laitinen, L. 1995. “Nollapersoona [The zero person].” Virittäjä 99: 337–357.

Lampinen, A. 1990. “Suomen kielen kohteliaisuusstrategiat [Politeness strategies in Finnish].”
In Sananjalka. Suomen kielen seuran vuosikirja 32 [Sananjalka. Yearbook of the Finnish
Language Society 32], A. Hakanen (ed.), 77–92. Turku: Finnish Language Society.
Luukka, M.-R. 1992a. Akateemista metadiskurssia. Tieteellisten tekstien tekstuaalisia, interper-
sonaalisia ja kontekstuaalisia piirteitä [Academic Metadiscourse. Textual, Interpersonal and
Contextual Features of Academic Texts]. Jyväskylä: University of Jyväskylä.
Luukka, M.-R. 1992b. “Varmuuden kahdet kasvot tieteellisessä tekstissä [The Janus-faced cer-
tainty in academic texts].” Virittäjä 2: 361–379.
Nikula, T. 1996. Pragmatic Force Modifiers. A Study in Interlanguage Pragmatics. Jyväskylä: Uni-
versity of Jyväskylä.
Ochs, E., Schegloff, E.A. and Thompson, S.A. (eds.). 1996. Interaction and Grammar. Cam-
Paunonen, H. 1995. “Puhesuomen muuttuva omistusmuotojärjestelmä [Morphological chang-
es in spoken Finnish possessive forms].” Virittäjä 4: 501–531.
Pomerantz, A. 1978. “Compliment responses: Notes on the co-operation of multiple con-
straints.” In Studies in the Organization of Conversational Interaction, J. Schenkein (ed.),
79–112. New York: Academic Press.
Pomerantz, A. 1984. “Agreeing and disagreeing with assessments: Some features of preferred/
dispreferred turn shapes.” In Structures of Social Action: Studies in Conversation Analysis,
Prince, E., Frader, J. and Bosk, C. 1982. “On hedging in physician-physician discourse.” In Lin-
guistics and the Professions: Proceedings of the Second Annual Delaware Symposium on Lan-
guage Studies, R.J. di Pietro (ed.), 83–97. Norwood, NJ: Ablex.
Sacks, H. 1992. Lectures on Conversation. Oxford: Blackwell.
Sacks, H., Schegloff, E.A. and Jefferson, G. 1974. “A simplest systematics for the organization of
turn-taking for conversation.” Language 50 (4): 696–735.
Schegloff, E.A. 1996. “Confirming allusions: An empirical account of action.” American Journal
of Sociology 102 (1): 161–216.
Scheibman, J. 2001. “Local patterns of subjectivity in person and verb type in American English
conversation.” In Frequency and the Emergence of Linguistic Structure, J. Bybee and P. Hop-
per (eds.), 61–89. Amsterdam: John Benjamins.
Selting, M. and Couper-Kuhlen, E. (eds.). 2001. Studies in Interactional Linguistics. Amsterdam:
John Benjamins.
Seppänen, E.-L. 1997. “Vuorovaikutus paperilla [Interaction on paper].” In Keskustelunanalyy-
sin perusteet [Introduction to Conversation Analysis], L. Tainio (ed.), 18–31. Tampere:
Vastapaino.
Sorjonen, M.-L. 1989. “Vuoronalkuiset konnektorit: mutta [Turn-initial connectors: mutta
‘but’].” In Kieli 4: Suomalaisen keskustelun keinoja I [Characteristics of Finnish Conversation
I], A. Hakulinen (ed.), 162–176. Helsinki: Department of Finnish, University of Helsinki.
Sorjonen, M.-L. 1999. “Dialogipartikkelien tehtävistä [On the use of response words].” Virit-
täjä 103: 170–194.
Sorjonen, M.-L. 2001. Responding in Conversation: A Study of Response Particles in Finnish.
Tainio, L. 1993. Kannanoton tulkinta keskustelussa [Interpreting an Assessment in Conversation].
Licentiate’s Thesis, Department of Finnish, University of Helsinki.
Tainio, L. 1996. “Kannanotoista arkikeskustelussa [On assessments in everyday conversation].”

In Kieli 10: Suomalaisen keskustelun keinoja II [Characteristics of Finnish Conversation II],
A. Hakulinen (ed.), 81–108. Department of Finnish, University of Helsinki.
Weber, E.G. and Bentivoglio, P. 1991. “Verbs of cognition in spoken Spanish: A discourse pro-
file.” In Discourse-Pragmatics and the Verb, S. Fleischman and L.R. Waugh (eds.), 194–213.
London: Routledge.
Wu, R.-J.R. 2004. Stance in Talk: A Conversation Analysis of Mandarin Final Particles. Amster-
Appendix A: Symbols used in transcription
(From Du Bois et al. 1992; Du Bois et al. 1993)
{carriage return} Intonation unit (IU)

-- Truncated intonation unit
- Truncated word
. Final intonation contour
, Continuing intonation contour
[ ] , [2 2] Speech overlap (indexed by numbers inside brackets)
= Lengthening
…(N) Long and medium pause (duration in parentheses; marked on a
separate line if not within an IU)
.. Short pause (brief break in speech rhythm)
(TSK), (SNIFF) Vocal noises
(Hx) Exhalation
(H) Inhalation
 Soft voice (piano)
<A> </A> Rapid speech (allegro)
<HI> </HI> Higher pitch level
<%> </%> Creaky voice
<X> </X> Uncertain hearing
X Indecipherable syllable
(( )) Researcher’s comment
Appendix B: Symbols used in glossing
1sg first-person singular ending

2sg second-person singular ending
2pl second-person plural ending
3sg third-person singular ending
252 Mirka Rauniomaa
4 passive person ending

abl ablative (‘from’)
acc accusative
ade adessive (‘at,’ ‘on’)
adv adverbial
all allative (‘to’)
cli clitic
comp comparative
cond conditional
ela elative (‘out of,’ ‘from’)
gen genitive
ill illative (‘into,’ ‘to’)
imp imperative
ine inessive (‘in’)
inf infinitive
neg negation
pass passive
pcp participle
pl plural
prt particle
ptv partitive (‘part of ’)
q question marker
trans translative (‘into,’ ‘for’)
Stancetaking as an interactional activity
Challenging the prior speaker
Tiina Keisanen
1. Introduction1
The study of how people display affect, evaluation, or epistemic certainty (or
doubt) toward some state of affairs, negotiate their points of view and alignment
with each other, or in short, the study of stancetaking, has gained more interest
lately within linguistics and also in other related areas of research. In order to
widen the scope of research from explicitly evaluative or epistemic material in
individual utterances, which has been the focus of much past linguistic research
on stance, I will here start from a position whereby language in interaction is
chosen as the domain of study, and stancetaking is seen as an intersubjective, so-
cially constructed activity (Du Bois this volume; see also Kärkkäinen 2003). More
specifically, stancetaking is conceptualized here through the social-interactional
processes of alignment and disalignment between discourse participants (cf. Du
Bois this volume). Within the conversation analytic framework used in the pres-
ent study, this is taken to mean that when a subsequent turn aligns with a prior
turn, it follows the projected course of action of the prior turn or the sequence
(cf. Keevallik 2003: 29). The absence of such alignment is considered as a display
of disalignment. This study discusses three types of polar interrogatives, namely
negative yes/no interrogatives and two types of tag questions, as doing disaligning
work, specifically challenging, in their local contexts. The challenging actions are
constructed by displaying doubt toward a claim or a stance embedded in the prior
turn, thereby suggesting that this stance or claim is problematic, and holding the
recipient accountable for it.
The present study is underlined by a certain view of language, a view that
is dominant especially among conversation analysts and interactional linguists:
all aspects of language are considered to be sensitive to the social action that the
254 Tiina Keisanen
discourse participants are engaged in at a given moment (Ford et al. 2003; Good-
win 2002; Schegloff 1995). Social-interactional practices are thus taken to form
the framework within which lexis, grammar, prosody, and all other features of
language acquire local meanings. In other words, not only is stance considered
to be an interactional achievement, but also the linguistic practices involved in it
are considered to be contingent products of interaction, that is, the practices of
stancetaking are examined in and through the sequential organization of interac-
tion and the actions and activities people are engaged in. Thus, I am supporting
a view that in some sequential contexts, any linguistic or paralinguistic feature
of language, or a linguistic construction for that matter, can function as a stance
marker. Before presenting the database in more detail, I will briefly discuss prior
interactional research on challenging as an interactional practice.
There are a small number of studies which have examined the features of chal-
lenging actions in interaction. Koshik’s (2003) study on challenging wh-interroga-
tives, such as ok how’s it background in the example below, is among these few.
(1) (Koshik 2003: 60–61)

1 TC: Is it relevant?
2 *to what you’re saying?
3 (0.8)* ((SD is looking down from * to *))
4 SD: No it’s just background. heh h
5 TC: It’s background?=
6 SD: =yeah.=
7 TC: ok how’s it background.=because u- like .h most
8 people wouldn’t know[:
9 SD: [((sniff))
10 TC: maybe what he meant by direct indirect
11 ordinary innovative or visionary.
12 SD: yeah. ok,
In a certain sequential context, discourse participants are observed to use the wh-
interrogative on line 7, and other wh-interrogatives like it, for challenging “the
basis for or right to do an action done by the prior utterance” (Koshik 2003: 51).
Consequently, challenging wh-interrogatives function as requests for an account
for the prior utterance. The challenges are not, however, designed to make a re-
sponse relevant. This is because the challenging wh-interrogatives are found to
convey a negative assertion, e.g., ok how’s it background conveys ‘It’s not back-
ground,’ which suggests that according to the speaker producing the wh-inter-
rogative the grounds for doing the prior action do not exist, and therefore, re-
sponding is not treated as the relevant next action (Koshik 2003: 63). Moreover,
Challenging the prior speaker 255
in Koshik’s study the wh-interrogatives used as challenges typically occur in an

already established environment of disagreement (e.g., after accusations) “where
challenging is a sequentially appropriate next response” (2003: 52). Additionally,
Heritage (2002) offers some observations on “argumentative challenges” in news
interviews. Interviewers regularly construct such turns by combining negative
yes/no interrogatives with critical question prefaces and/or other evaluative mate-
rials. However, like the wh-challenges discussed in Koshik (2003), the interview-
ers’ turns are rendered essentially as unanswerable assertions due to the fact that
they prefer responses that invite the interviewees to assent to critical or negative
evaluations of their own conduct or that of their allies. The interviewees are also
observed to treat the interviewer’s turns as hostile assertions since they typically
respond with disagreements and rebuttals. This paper adds to these prior stud-
ies by examining challenging negative yes/no interrogatives as well as tag ques-
tions in two different sequential contexts in everyday interactions. It is argued
that the challenges discussed here are warranted by a discrepancy in information
or knowledge between the participants, and thereby relate to the social co-or-
dination of knowledge, rather than being presented as unanswerable or hostile
assertions.
The study is organized in the following manner. Section 2 introduces the da-
tabase used in the study, while Section 3 presents the main findings. The focus will
be on explicating two sequential contexts of challenging negative yes/no inter-
rogatives and tag questions in the current database. It is argued that in their local
contexts the interrogative turns are observed to constitute a disaligning challenge
in the unfolding course of action by calling into question a claim or a position
embedded in the prior turn, and requesting that the recipient accounts for this
observed discrepancy between the participants. Section 4 concludes the study
with a discussion of the findings.
2. Data
The data for this study come from the first three volumes of the Santa Barbara
Corpus of Spoken American English (SBCSAE) (Du Bois et al. 2000; Du Bois et
al. 2003; Du Bois and Englebretson 2004). The database includes mostly face-to-
face conversations in everyday settings, but also some task-oriented interactions
such as workplace talk and service encounters. The data are drawn from 42 speech
events, which total about 16 hours of transcribed speech.
The selection of data was initially based on morphosyntactic criteria, i.e., I
searched the corpus for utterances with both negation and inverted word order
(that is, one where a verb, either the auxiliary/operator or the main verb precedes
256 Tiina Keisanen
the subject). Most such structures are polar interrogatives, either negative yes/no
interrogatives such as, Isn’t that an oil tank?, or tag question constructions such
as, you haven’t .. r- really lived in the house .. during the winter, have you? (nega-
tive declarative followed by a positive tag) or .. It’s kind of smelly, isn’t it. (posi-
tive declarative followed by a negative tag). These three constructions share the
general characteristic with other interrogative forms in that they regularly make
a response relevant when used in conversation (Thompson 1998). As such, inter-
rogatives are often found to initiate “adjacency pair” sequences, which, according
to Schegloff and Sacks (1973) consist of two relatively ordered, type-connected
turns in which a certain type of first pair part makes a choice within a set of re-
sponse types “conditionally relevant” (Schegloff 1968). Adjacency-pair sequences
provide the discourse participants with a basic resource for the organization of
action in social interaction. Moreover, all three interrogative constructions ex-
amined here are subtypes of polar, or yes/no, interrogatives, whose grammati-
cal form constrains the recipient’s response to two alternatives, i.e., the recipient
is guided to choose between a yes or no response (Raymond 2003). The three
constructions therefore share many (grammatical) characteristics. However, this
should not be taken to imply that they are considered to be identical grammati-
cally or pragmatically. For example, prior linguistic research has argued that the
constructions embody differences in presupposition, and consequently set up dif-
ferent trajectories for the recipient’s response in terms of its polarity (this has
been called “conduciveness” or “epistemic bias” in the literature; see, for example,
Givón 2001; Quirk et al. 1985; Sadock and Zwicky 1985). On the other hand, it
should be noted that in interactional research, concepts such as conduciveness,
and relatedly, epistemic strength of a turn, should be seen as empirical problems to
be worked out from the specifics of the local context in which a turn is produced,
rather than basing such judgments exclusively on form or intuition, for instance.
Koshik (2005) illustrates this by discussing two contrastive accounts of the epis-
temic strength of negative yes/no interrogatives in interaction. The first comes
from news interviews. As Heritage (2002) has observed, in news interviews nega-
tive yes/no interrogatives are regularly used to implement hostile assertions of
opinion, which may be agreed or disagreed with, rather than being presented as
“questions” to be answered. On the other hand, tag questions (in this case posi-
tive statements followed by negative tags, since negative statements followed by
positive tags were not discussed in Heritage’s study) were found to function more
like “yes/no questions to be answered” in news interviews. In other words, in the
news interview context negative yes/no interrogatives are presented as conducive,
epistemically strong assertions, while tag questions are not. On the other hand,
Koshik (2005) discusses an example from an everyday interaction (originally pre-
sented in Schegloff [1995] 2007). In that case, a negative yes/no interrogative is
observed to weaken the epistemic strength of a prior tag question in the face of
incipient disagreement. Example (3) below presents one sequentially very similar
case from the current database.
However, instead of examining the conduciveness or the relative epistemic
strength between the chosen constructions in more detail in everyday interac-
tions and/or in different sequential contexts, this paper concentrates on providing
an analysis of the interactional and sequential organization of action at points in
which negative yes/no interrogatives and tag questions are found to function as
disaligning challenges in the unfolding course of action. Moreover, epistemicity is
here used to refer to those interactional and linguistic means by which discourse
participants display their certainty or doubt toward some state of affairs or a piece
of information in their own turn, or in the turns of others.
171 instances of tag questions and negative yes/no interrogatives found in the
SBCSAE form the database used in this study. Table 1 below provides informa-
tion on the frequency of different constructions and their functional distribution
in the database.
Requesting confirmation or agreement was found to be the most common
functional category for both tag questions and negative yes/no interrogatives in
the current database. Negative yes/no interrogatives were also found rather fre-
quently to do assessing. The category other includes turns in which negative yes/no
interrogatives and tag questions carry out some less frequent interactional func-
tions such as requesting or reporting, or they are produced during a monologue
(e.g., in a lecture or a sermon) or as reported speech (typically in the midst of
storytelling sequences). In this paper, I will concentrate on negative yes/no inter-
rogatives and tag questions that are used as challenges (see Keisanen 2006 for a
discussion of the other data). The data consist of 15 tag questions and 21 negative
yes/no interrogatives.
The data are transcribed according to the transcription system outlined in
Du Bois et al. (1993). I have slightly modified the original transcripts in order to
Table 1. Tag questions and negative yes/no interrogatives in the database by function
Tag questions Negative yes/no interrogatives
Tokens Percentage Tokens Percentage
Requesting confirmation or 31 44.9% 31 30.4%
agreement
Assessments 12 17.4% 28 27.4%
Challenges 15 21.7% 21 20.6%
Other 11 15.9% 22 21.6%
Total 69 100% 102 100%
258 Tiina Keisanen
better demonstrate the contribution of prosody and timing for the accomplish-
ment of disaligning challenges. First, more prosodic detail is added especially on
the interrogative turns and on turns adjacent to them. Second, to be more in line
with the standard conversation analytic transcription practice, the placement of
many of the longer pauses has been changed from the beginning of an intonation
unit to a line of their own. A list of symbols used in the transcripts can be found
in the Appendix.
3. Doing challenging with negative yes/no interrogatives

and tag questions
This section discusses two sequential contexts of challenging tag questions and
negative yes/no interrogatives in the current database. Section 3.1 includes cases
of negative yes/no interrogatives and tag questions that are responsive to types of
telling sequences, while Section 3.2 concentrates on sequences where the con-
structions are used as responses to some initiating actions.
3.1 Challenging the trajectory of an extended sequence
The challenging negative yes/no interrogatives and tag questions discussed in this
section are responsive to types of personal accounts such as stories, tellings, and
the like. As part of such multi-unit accounts, participants typically make various
types of claims and evaluations, as well as design their tellings so as to seek af-
filiation or a certain type of stance display from the recipient(s) (Clift 2000; Ford
2004). The challenging negative yes/no interrogatives and tag questions that are
responsive to extended sequences draw attention to some (explicit or implicit)
claim made or a stance taken in the telling by exploiting a place in the sequence in
which it would be appropriate to align with the teller, but in which a response is
not actually prompted by the teller. By calling into question a claim or a position
embedded in the prior speaker’s turn, the interrogative speaker holds the prior
speaker accountable for this position or claim. Simultaneously, the interrogatives
constitute a display of disalignment toward the unfolding course of action. The
following examples illustrate and discuss this use of negative yes/no interrogatives
and tag questions in the database.
The first excerpt below has three participants, Kathy, Sharon and Carolyn,
who are sisters. Just prior to the excerpt, Sharon has been talking about her expe-
riences as a substitute teacher. She has been complaining about how a lot of the
children in her mixed class of third and fourth graders have trouble with the times
table. Sharon blames their previous teachers for being bad teachers: they had not
taught them times tables in the second grade, which she thinks should have hap-
pened. Carolyn’s turn that begins on line 2 responds to Sharon’s complaint.
(2) (Raging Bureaucracy SBCSAE 0004 <T: 00.18.31>)

1 KATHY: The teachers have just8] been [9passing them along XXX9].
2 CAROLYN: [9Wait a minute,
3 neither did I=9].
4 SHARON: [Yeah].
5 CAROLYN: [I] didn’t know them either.
6 ... I didn’t know times tables till s- --
7 .. like fifth gra=de.
8 Only cause I was a dufus.
9 (H) But,
10 SHARON: ... [Well <X were X>] --
11 KATHY: [No,
12 <X It was] [2because you were a X> lazy butt2].
13 CAROLYN: [2I could not --
14 I= could not2] do fractions,
15 .. I could not do fractions.
16 .. And I mean with .. ^~Samuel as a `teacher,
17 who ^wants= to ^tell <X them X>,
18 .. (H) who `wants to go up <@ to this ^man and say @>,
19 <VOX Ì have a ^problem VOX>.
20 (0.3)
21 CAROLYN: (GASP)[=]
22 KATHY: [<HI You] `didn’t HI> really ^have Mister Samuel,
23 ^did you?
24 (0.3)
25 CAROLYN: Ôh=,
26 hell ^yeah.
27 [@]
28 KATHY: [Ôh],
29 I `thought you got ôut of being in his `cla[ss].
30 CAROLYN: [Ùnh]-ùnh=.
31 I ^had him,
32 .. my ^whole fourth-`grade year.
Carolyn’s telling of her troubles with learning math at school starts in overlap with
Kathy on line 2. By demonstrating through her own experiences that difficulties
with math are common in third and fourth grade, Carolyn’s turn downplays Sha-
260 Tiina Keisanen
ron’s complaint about the students and their teachers at Sharon’s school. In effect,
Carolyn provides evidence that Sharon’s expectations may be unrealistic. In con-
junction with this, Carolyn brings in her old teacher, Mr. Samuel, by saying, And I
mean with .. Samuel as a teacher, who wants= to tell <X them X>, who wants to go
up to this man and say, <VOX I have a problem VOX> (lines 16–19). Mr Samuel’s
violent teaching methods were talked about earlier; here, that is implicitly referred
to as one reason for why Carolyn did not learn certain things at the same time as
others. Carolyn’s extended turn ends with an emphatic demonstration of her fear
of Mr. Samuel, <VOX Ì have a ^problem VOX>, (0.3) (GASP)= (lines 19–21),
which imitates the voice of a little girl, and which is added with a gasping sound
that imitates a small girl being scared. In other words, Carolyn designs these in-
tonation units by using direct reported speech and altered voice quality (the por-
tion of the transcript marked with VOX), thereby indexing this as the climax of
the telling (Clift 2000; Drew 1998). Clift (2000) observes that such climaxes seek
affiliation from the recipient, but in an implicit manner, that is, a response is not
actually solicited. Instead of affiliating with Carolyn’s telling as would be relevant,
however, Kathy disaligns with the projected course of action by producing a tag
question challenge2 <HI You `didn’t HI> really ^have Mister Samuel, ^did you?
(lines 22–23). It singles out something that Kathy considers to be discrepant in
the content of Carolyn’s telling, namely the claim that Mister Samuel taught her at
elementary school. In so doing, Kathy displays doubt toward the accuracy of this
claim, thereby holding Carolyn accountable for the discrepancy.
The tag question includes a component that displays Kathy’s affective stance:
her prosody is marked to show emotion, which can vernacularly be called surprise
or the like. This heightens the degree of doubt displayed with the tag question.
Figure 1 below shows the pitch trace for the tag question.
The main prosodic feature that displays affect is the overall high pitch during
the turn, and the extremely high pitch in the beginning of the turn. The high-ris-
ing final intonation also contributes to the affective quality of the tag question. In
sum, the display of affect provides further warrant for the challenge embedded in
the tag question turn. Moreover, both parts of the verb phrase didn’t and have on
line 22 receive accent. In conjunction with the adverbial really, they mark contrast
with the claim that Mr. Samuel was teaching Carolyn; this claim was embedded in
Carolyn’s turn And I mean with ..^Samuel as a `teacher (line 16).
Carolyn’s prosodically and lexically emphatic response Ôh=, hell ^yeah
(lines 25–26) provides an example of a direct, categorical disagreement toward a
disaligning challenge carried out by a prior interrogative turn. It begins with turn
initial Ôh=, which in responses to prior inquiries mark that the inquiry “was
unexpected, unlooked for or ‘out of left field’” (Heritage 1998: 294). Nevertheless,
this is not enough to close the sequence. Kathy’s third-position turn (lines 28–29)
600
500
400
300
200
100
You didn't really have Mr Samuel did you
0 2.07667
Time (s)
Figure 1. Pitch trace of You didn’t really have Mr ~Samuel, did you?
begins with a turn-initial oh-token. In this context, in third position, it indicates

that she has gone through some change in her state of knowledge or orientation
(Heritage 1984). It is followed by an account for why she brought up the issue
in the first place, I `thought you got ôut of being in his `class. However, no cor-
rection to Kathy’s inaccurate understanding of Carolyn’s prior teachers is offered
yet. Carolyn therefore begins such an account on line 32, and the original topic is
never returned to.
In example (3) below, Michael is telling Jim how he would like to be the per-
son whose voice a telephone company has used for the recorded messages one
hears when calling a wrong number. The motivation behind Michael’s wish is the
money that comes from the royalties from the use of the voice.
(3) (Wonderful Abstract Notions SBCSAE 0016 <T: 00.01.11>)

1 MICHAEL: You know I wish I was,
2 uh,
3 the person=,
4 .. whose voice they used in the .. telephone,
5 when it tells you,
6 (0.4) the number has been changed,
7 .. (H) and that I had uh,
8 ^copyright,
9 @@
10 get some royalties.
11 Like,
12 (H) that ^lady,
13 you know,
262 Tiina Keisanen
14 h- you ^hear,
15 (H) (0.4) <Q the ^number you have `reached,
16 da-duh da-duh [da-duh da-duh da-duh da-duh Q>].
17 JIM: [They ^got a `different ^woman],
18 ^didn’t they.
19 (0.3)
20 MICHAEL: ^Hunh?
21 JIM: ^Didn’t they get a `different ^woman?
22 (0.2)
23 JIM: `When she th- `tried to do ^that?
24 MICHAEL: [I don’t `know if they ^did].
25 JIM: [They,
26 they,
27 they `hired an]other ^voi[2ce2].
28 MICHAEL: [2I2] didn’t `hear any ^follow-up.
29 But,
30 JIM: That’s what I thought that they did any[ways].
31 MICHAEL: [Yeah]?
32 (0.4)
33 MICHAEL: Hunh.
34 (1.4)
35 MICHAEL: They certainly u=sed her a lot.
Michael moves into demonstrating what the recorded messages sound like on
lines 14–16, h- you ^hear, (H) (0.4) <VOX the ^number you have `reached, da-duh
da-duh da-duh da-duh da-duh da-duh VOX>. As in example (2), these intonation
units are designed by using direct reported speech and an altered voice quality. At
this juncture of the telling, it would therefore be appropriate for Jim to align with
Michael’s telling by affiliating with it (e.g., by laughing, offering an assessment,
or some other agreeing or appreciative turn). However, instead of this, Jim’s tag-
question turn, They `got a different ^woman, ^didn’t they (lines 17–18), offers an
observation of a discrepancy in the information content of Michael’s telling. This
observation provides him warrant to come in at this point in the telling. Accord-
ing to Jim, the phone company fired the woman that Michael is talking about, be-
cause she tried to get money from the use of her voice. Jim’s tag question therefore
undermines Michael’s story by pointing out that someone already tried, but failed
in carrying out the wishful scheme that Michael is laying out here. Simultaneous-
ly, the tag turn constitutes a display of doubt (to the degree of dismissal) toward
Jim’s knowledge state with respect one of the presupposed facts of the telling (i.e.,
that the lady is still working for the company).
After a short pause (line 19), Michael responds to the tag question with a
repair initiation ^Hunh? (line 20). The repair initiation may be caused partly be-
cause of troubles in hearing due to the overlap on lines 16–17. On the other hand,
hunh belongs to the class of repair initiators that have been observed on occasion
to indicate troubles in the appropriateness, relevance, and/or the sequential fit
of the prior turn (Drew 1997). Thus, ^Hunh? may display that Michael does not
understand how the tag turn is connected to his telling, or further, that he does
not consider it to be related to his telling thereby pre-monitoring disagreement. In
the repair on lines 21–23, Jim first repeats the candidate claim, but reformulates it
as a negative yes/no interrogative, ^Didn’t they get a `different ^woman? (line 21),
and adds, `When she th- `tried to do ^that? (line 23), in order to explicate how the
interrogative constructions were connected to the telling. In this context, facing
potential disagreement, the reformulation of the tag question as a negative yes/no
interrogative can be seen to downgrade the epistemic strength of the tag question,
and therefore to indicate backdown from the challenge. Also the prosody reflects
a change in epistemic strength as it changes from a tag question produced with
final falling intonation, to a negative yes/no interrogative produced with rising-
query intonation.
In any case, Michael responds by claiming to have insufficient knowledge of
the situation Jim is referring to with I don’t `know if they ^did (line 24). Such a re-
sponse enables Michael to avoid confirming Jim’s inquiry on the factual level, and
more importantly, to reject the action carried out with the prior turn (see Beach
and Metzger 1997). Michael continues with an account, I didn’t `hear any ^fol-
low-up (line 28), which demonstrates that his lack of knowledge is not a matter of
just claiming not to know, but that this claim is based on not having any evidence
that would support Jim’s turn. Jim responds to this with an account of his prior
understanding, That’s what I thought that they did anyways (line 30). Even though
Jim uses past tense to display some change and/or uncertainty toward his position
(cf. example 2), he does not offer further evidence (e.g., the source of knowledge)
for his challenging turn. The matter is thus left somewhat unresolved.
The following excerpt contains another example of a tag question that is re-
sponsive to an extended sequence. Throughout the excerpt, Wendy and Marci are
involved in doing disaligning actions, which in part contributes to the interpre-
tation of the tag question on lines 20–22 as a challenge. The excerpt starts with
Marci’s complaint about the bad quality of the containers of the soda that they are
drinking.
264 Tiina Keisanen
(4) (Appease the Monster SBCSAE 0013 <T: 00.00.07>)

1 MARCI: .. Why do these ca=ns,
2 .. get so warped.
3 Only the --
4 .. Only the Sam’s Club cans .. get so [warped].
5 KEN: [(H)]=
6 WENDY: (H) % [2I think you should2] [3write3] --
7 KEN: [2<X Cause they’re X>2] [3X3] --
8 KENDRA: [3Cause they’re3] [4cheap ca=ns,
9 WENDY: [4%Just ..
%= write a letter to em4],
10 KENDRA: that’s why4] --
11 WENDY: and complain !Marci.
12 MARCI: .. (H)= `What I’m gonna ^complain about,
13 is that they ^don’t make white `grape.
14 ^This stuff [is] ^g=ood.
15 >ENV: [((POP-TOP))]
16 MARCI: It’s like ^sparkling= `grape `juice .. ^cocktail,
17 or `something,
18 .. you know,
19 .. `remember that [white] --
20 WENDY: [(H) <HI<F They] ônly make `that
F>HI>.. with ^Nutrasweet though,
21 <X do- X> --
22 .. ^don’t they.
23 .. They don’t `make (0.4) a ^regular --
24 (0.3)
25 MARCI: A `regular ^grape?
26 I don’t [^know].
27 KEN: [(THROAT)]=
28 (0.4)
29 WENDY: Every `time I’ve ^looked at <X the X> bot- that bottled ^water,
30 .. or the `sparkling ^water,
31 [it’s all] [2with2] --
32 MARCI: [But they don’t] [2make2] `this kind at â=ll [3any3]more.
33 KENDRA: [2(Hx)2]
34 [3(SNIFF)3]
35 [4(THROAT)4]
36 WENDY: [4.. Oh they4] don’t?
37 MARCI: .. I couldn’t find it last time I was in there.
Ken, Kendra, and Wendy react to Marci’s complaint on lines 1–4. Kendra offers
the cheapness of the soda as an account for the bad quality of the containers (line
8), thus implicating that warping is what can be expected from them. Ken also
starts to formulate an account with <X Cause they’re X> (line 7), but cuts off his
turn most likely due to the extended overlap first by Wendy and then Kendra. Ac-
counts are one type of relevant second pair part after complaints (Schegloff 1988).
However, Wendy disaffiliates with the complaint by producing a rather dismissive
turn, %Just .. %= write a letter to em, and complain !Marci (lines 9, 11). This turn
is produced in a somewhat sarcastic tone, which further strengthens the negative
character of the turn.
Wendy’s turn receives a counter from Marci in the form of a wh-cleft, `What
I’m gonna ^complain about, is that they ^don’t make white `grape (lines 12–13).
Wh-cleft constructions have been associated with stance displays (Hopper 2000;
Kim 1995), and specifically with disjunctive actions in which the wh-cleft may be
used to evoke the speaker’s first-hand epistemic experience of the matter (Kim
1995). This functions to “facilitate the foregrounding of the speaker’s counterac-
tive stance towards the preceding talk” (Kim 1995: 268). Thus, Marci asserts her
right to complain by using the cleft construction. After that, she moves on to ar-
gue for the good quality of the drink with ^This stuff is ^g=ood. It’s like `sparkling=
`grape `juice .. ^cocktail, (lines 14, 16). However, her assessment of the drink does
not get taken up immediately, and so Marci continues the turn with .. you know, ..
`remember that white -- (lines 18–19). Marci may be working her way into a story
or a reminiscence of a shared experience relating to the drink that would provide
more evidence of its good quality. However, whatever activity is started, it is inter-
rupted by Wendy’s tag question, (H) <HI<F They ônly make `that F>HI>.. with
^Nutrasweet though, <X do- X> -- ^don’t they (lines 20–22). In other words, in-
stead of joining Marci’s evaluation of the drink, or waiting to align as a recipient to
a story, Wendy latches on with a turn that questions the quality or the healthiness
of the drink (Nutrasweet is an artificial sweetener, which is believed by some to be
toxic). The main accent on line 20 is on the adverbial only, which here functions
as a marker that restricts a claim made in previous discourse, and carries nega-
tive connotations toward the use of Nutrasweet. Essentially, then, by appealing to
the potential unhealthiness of the drink, Wendy disaligns with the course of ac-
tion that Marci pursues. Moreover, the challenge carried out with the turn makes
Marci’s positively evaluative stance of the beverage an accountable action.
Marci’s response to the tag question on lines 25–26 consists of two parts. A
`regular ^grape (line 25) is a candidate offered for Wendy’s word search, They
don’t `make (0.4) a ^regular -- (line 23). The second part of the turn, I don’t ^know,
(line 26), contains Marci’s response to Wendy’s claim that the company does not
make regular soda any more. As in example (3), I don’t ^know, does not confirm
266 Tiina Keisanen
Wendy’s inquiry on the factual level. Moreover, it enables her to avoid conceding
to the challenge carried out with the tag-question turn.
After the challenge is met with a non-committing and disaffiliative response,
Wendy moves on to provide first-hand evidence for her tag-question turn with
Every `time I’ve ^looked at <X the X> bot- that bottled ^water, (line 29). As ob-
served by Pomerantz (1984a), this typically takes place if other discourse par-
ticipants display doubt toward the validity of other participants’ claims. However,
Wendy does not get a chance to finish her turn, because Marci’s disagreeing turn,
But they don’t make `this kind at â=ll anymore (line 32) starts in overlap with
hers. The turn includes the emphatic at âll=, which indicates that whether or
not the company uses Nutrasweet is not really the issue. In short, by pointing out
a flaw in Wendy’s account and consequently, in her challenging turn, Marci reas-
serts her right to complain about the state of affairs, and maintains her positively
evaluative position toward the drink.
The following example presents a sequential variation to the disaligning nega-
tive yes/no interrogatives and tag questions that are responsive to types of telling
sequences. In these cases, the negative yes/no interrogative or a tag question is
preceded by a repair sequence, which concerns the same issue as the following in-
terrogative. In other words, the interrogative is produced after the repair sequence
has not successfully solved a problem one of the recipients observes in the telling.
Consequently, in such cases, the repair sequence is employed as pre-disagreement
(Schegloff et al. 1977), while the following negative interrogative carries out the
fully formulated disagreement, or challenge. Koshik’s (2003) discussion on wh-
interrogative challenges includes a similar sequence from a teacher-student tutor-
ing session. In the extract below, Marie is telling Lisa and Kevin about taking her
son Issac to the doctor. Lisa and Marie are close friends, Kevin is Lisa’s brother.
(5) (Judgmental on people SBCSAE 0036 <T: 00.06.05>)

1 MARIE: [the next time I took him] he’s all,
2 (H) (SCREAM)
3 [I was all <VOX oh my] ^Go=d VOX>.
4 KEVIN: [@@@@]
5 MARIE: And [2then I pick2]ed him up,
6 LISA: [2XX2]
7 MARIE: and his ^legs were ^bleeding and stuff,
8 and I had ^blood all on my `shirt,
9 and I was all <VOX ôh my @^Go=d VOX> [@@@(H)].
10 LISA: [^Blood- %(Hx)] --
11 MARIE: (H) ^Blood,
12 cause he ^bleeds.
13 (0.5)
14 (TSK) (H) After they give him a ^shot?
15 KEVIN: But ^don’t they,
16 like [put a ^thing on it right àf]ter?
17 LISA: [But that ^mu=ch=]?
18 (0.4)
19 MARIE: ^No,
20 but it was just coming ôut.
21 Do you know what I mean,
22 it was like,
23 [mm],
24 KEVIN: [<HI Well they] should have HI> put>,
25 I mean ûsual[2ly you know like when you like2] [3give ^bloo=d,
26 MARIE: [2A ^bandaid2]?
27 LISA: [3But it wasn’t âll= ov3]er,
28 KEVIN: they use3],
29 LISA: it was just like=,
30 .. a [little ^dr]ibble.
31 MARIE: [It had] --
32 N_yeah no,
33 it had .. `gotten on my ^shirt,
34 and I was [just all] <VOX òh my ^Go=d,
35 LISA: [Oh],
36 MARIE: (0.4) my child’s `bleeding VOX>.
The telling that starts on line 1 follows an account of a no-problem visit to the doc-
tor. According to Marie, Issac was crying hard after the shot, in addition to which
he was bleeding. Unlike earlier, where the first part of the story was received with
recipient appreciation, (H) That’s good (data not shown), this part of the story
gives rise to a trouble source: on line 10 Lisa initiates a repair sequence by produc-
ing a partial repeat of the prior turn, ^Blood- %(Hx) --. Marie repairs the trouble
by first reproducing the word blood (line 11), and then supplying cause he ^bleeds
(line 12) by way of an explanation for why she would have blood on her shirt. This
type of repair would be sufficient to repair a trouble in hearing or understanding,
but as is pre-monitored by the pause on line 13, the trouble lies elsewhere. Marie
therefore continues to expand the repair turn by complementing it with a further
turn constructional unit, (TSK) (H) After they give him a ^shot? (line 14). It of-
fers an explication for why Issac would be bleeding in the first place. However,
first Kevin (lines 15–16) and then Lisa (line 17) both move on to produce further
268 Tiina Keisanen
turns, which in slightly differing ways challenge Marie’s telling about her son’s
excessive bleeding after receiving a shot.
Kevin’s turn, But ^don’t they, like put a ^thing on it right àfter? (lines 15–16),
begins with the contrast marker but, which is frequently used in the beginning of
disagreeing turns (Pomerantz 1984b). It is followed by a negative yes/no interrog-
ative, which questions Marie’s account of the procedure that her son underwent.
The challenge is formulated as a generalization and grounded on knowledge that
Kevin has acquired through his prior experiences with doctors and needles (as is
evidenced by his turn on lines 24–25, 28). According to Kevin, the spot should
be covered with a bandaid or the like immediately after the needle is pulled out.
In short, Kevin’s interrogative displays doubt toward Marie’s telling by suggesting
that Marie presents an inaccurate or a misleading description of what actually
happened. In so doing, the negative yes/no interrogative is used as a request for
an account for this discrepancy in information or knowledge between the par-
ticipants.
However, before Kevin finishes his turn, Lisa latches on in overlap with her
disagreeing turn, But that ^mu=ch=? (line 17). It also challenges Marie’s descrip-
tion of the excessive bleeding. There are thus two challenges on the table, which
cause some trouble for responding. Marie’s response, No, but it was just coming
out. Do you know what I mean, it was like, mm, (lines 19–23), is slightly ambiva-
lent as to whom she is responding. In any case, it begins with the negative item,
No, which indicates disagreement. The remainder of her turn can be seen to pro-
vide a slight correction, and a back down, to her telling. Marie does not change
any of the facts of her telling, however, she only specifies its timeline by claiming
that everything happened so fast (i.e., it was just coming out) that the doctor did
not have time to cover the spot before the blood was already all over her shirt.
Kevin overlaps the last intonation unit of Marie’s response with <HI Well they
should have HI> put>, (line 24). This intonation unit begins his account of how
the procedure should have gone. Additionally, Lisa tries to reformulate a detail
in Marie’s telling by producing, But it wasn’t âll= over, it was just like=, .. a little
^dribble (lines 27, 29–30). However, Marie does not back down any further. She
disagrees with Marie (line 32) and goes on to redo the climax of the story (lines
33–34, 36), which was interrupted by the other-initiated repair and challenges
discussed above. In effect, Marie restates her position and disregards both Kevin
and Lisa’s challenges to her claims.
In sum, the challenging negative yes/no interrogatives and tag questions dis-
cussed above provide a resource for story recipients with which they can disalign
with the tellers and the projected trajectory of action. The sequential placement is
the decisive factor in the interpretation of the interrogatives as disaligning chal-
lenges: the interrogatives are triggered by the prior turn, but they do not consti-
tute a sequentially relevant next action projected by the current activity, such as
alignment as a story recipient who would affiliate with the telling. The challenges
are based on an observed discrepancy between the participants with respect to
some detail in the telling. More specifically, the discrepancy is used to call into
question a claim or a position embedded in the prior turn, and moreover, to hold
the recipient accountable for these claims or stances. The recipients may counter
the challenges with explicit disagreement (example 2), or avoid accountability by
appealing to insufficient knowledge (examples 3 and 4). However, recipients were
also seen to provide accounts (example 2), or to back down from their claims (ex-
ample 5). Additionally, the interrogative speakers may provide accounts for their
turns (examples 4 and 5). Such responses provide evidence for the disaligning
nature of the interrogatives discussed above.
Negative yes/no interrogatives and tag questions that are used to challenge
the prior turn by pointing out a discrepancy in its content are also found outside
extended sequences. The following section discusses challenging negative yes/no
interrogatives and tag questions that provide a disaligning response to some ini-
tiating action.
3.2 Challenging prior action
Similarly to Section 3.1, the challenges discussed in this section are constructed
by pointing toward a discrepancy between the participants with respect to some
claim or position embedded in the prior turn, thereby calling this claim or posi-
tion into question. However, the negative yes/no interrogatives and tag questions
discussed in this section provide a disaligning turn to some initiating action, and
are used to challenge the appropriateness or the relevance of doing the action
completed in the prior turn. Such challenges bear a resemblance to wh-question
challenges discussed in Koshik (2003), which were observed to challenge the ba-
sis or right for doing the prior action. In the first example below, Alina and Lenore
are talking about Alina’s husband, Hector.
(6) (Cuz SBCSAE 0006 <T:00.20.34>)

1 ALINA: (TSK) `Poor little ^~Hector.
2 .. `He’s been `having a `hell of a ^time.
3 .. `Pulling his `hair out.
4 (H) ^He’s been `grumpy for `three months.
5 (H)
6 LENORE: <HI `Well,
7 it’s ô[ver now,
270 Tiina Keisanen
8 ALINA: [(H)] (SIGH)

9 LENORE: ìsn’t it HI>]?
10 ALINA: <HI ^Yeah HI>.
11 .. He was âctually ^talkative `tonight.
12 When I `called him on the ^phone?
13 .. (H) `He ^tal=ked to me.
14 For like `fifteen ^minutes tonight.
15 (H) Do you know how long it’s `been since he’s been able to
k- ^talk to me on [the `phone]?
16 LENORE: [`Who does he ^work] `with [2`now2].
17 ALINA: [2(H)2]
18 (TSK) [3!^Kevin3].
19 LENORE: [3!^Rosen3]blu[4m4]?
In the beginning of the excerpt, Alina initiates a new topic after talking at length
about a party she went to recently. On lines 1–4, she tells Lenore that her husband
Hector (who is not present) is going through a difficult period at work. Her turn
includes a display of an evaluative stance toward the informing. First, it includes
the negatively evaluative expressions hell of a time and grumpy. Second, her pros-
ody is marked to display that the first three intonation units deliver ‘bad news’:
the pitch level is low, the voice quality creaky, and the pitch range rather narrow
(Freese and Maynard 1998). However, in the last intonation unit on line 4, Alina
moves from characterizing her husband as poor little Hector (line 1) to grumpy
(line 4). The prosody also changes, the turn begins in rather high pitch, and the
pitch range is wider. In other words, Alina moves to position Hector’s situation
with respect to herself. One turn design feature that points toward this is the ad-
jective grumpy, which characterizes Hector’s behavior toward Alina. The last part
of the turn therefore moves more toward complaining than delivering news.
If Alina’s turn were taken up as a complaint (or alternatively, as an inform-
ing), an aligning response could consist of a co-complaint, an agreement, or other
alignment with respect to its status as a complaint (Schegloff 1988). However, Le-
nore’s turn does not provide such alignment. She points out with the tag question,
`Well, it’s ôver now, ìsn’t it? (lines 6–7, 9), that Alina’s prior turn includes a claim
or a position that requires correction, specifically by making available a contras-
tive formulation of the state of affairs at Hector’s work (Alina has described the
situation as an ongoing one, while Lenore depicts it to be over).3
Lenore’s tag question is designed by using high pitch and high-rising final
intonation. Figure 2 presents the pitch trace for the tag question.
The use of marked prosody adds affect to the turn, and provides further war-
rant for the challenge.
Figure 2. Pitch trace of Well it’s over now, isn’t it?
Lenore disaligns with the projected trajectory of the complaint sequence by

calling into question a claim embedded in Alina’s complaint. On the other hand,
it could be noted that the actual content of the challenging turn can be seen to
display compassion or sympathy: Lenore’s tag question turn indicates that the
situation is not as bad as Alina claims it to be, which could be taken also as a posi-
tive thing. So, to put it differently, by calling into question Alina’s formulation of
the situation, Lenore challenges the appropriateness of doing a complaint in the
first place. As can be seen from Alina’s response (lines 10–15), the challenge con-
tributes to a complete change in the direction of talk: Alina moves on to provide a
positively evaluative account of Hector and his behavior toward her.
The example below presents another case where the interrogative speaker
challenges the appropriateness of the prior action. It is done by targeting an em-
bedded claim (i.e., a presupposition) in the prior turn. The participants are cel-
ebrating Kendra’s birthday. Marci is Kendra’s mother, and has bought the cake
that has been brought in just prior to the excerpt. The others have also sung Happy
Birthday to Kendra.
(7) (Appease the Monster SBCSAE 0013 <T: 00:18:24>)

1 KENDRA: It’s a ^beautiful `cake,
2 but `why do you guys [5àlways5] give me îce cream `cakes.
3 WENDY: [5(THROAT)5]
4 (0.4)
5 KEVIN: Because [6it’s the ônly kind we’re `not6] [7àllergic to7].
6 MARCI: [6<HI Don’t you ^like ice `creamHI>6]?
7 KENDRA: [7I don’t `like ice7]
^crea=m.
272 Tiina Keisanen
8 (0.3)
9 WENDY: Do you like .. ^frozen `yogurt?
10 (0.6)
11 KENDRA: Î shouldn’t `blow this òut.
12 .. `Dad,
13 do you wanna ^try <X it X> for `me?
Kendra’s turn on lines 1–2 amounts to a complaint, even though the first intona-
tion unit, It’s a ^beautiful `cake, (line 1), compliments the cake, thus taking off the
sharpest edge from the complaint. Both Kevin and Marci offer their response to
Kendra’s complaint. On the level of form, Kevin’s response can be seen to provide
an aligning response to the complaint as it offers an account (line 5). However, on
the level of content Kevin’s turn counters the complaint with rather bold sarcasm.
Marci, in turn, disaligns with the complaint with a negative yes/no interrogative,
Don’t you ^like ice `cream? (line 6), which targets the basis of Kendra’s complaint.
It displays that Marci did not expect Kendra not to like the ice-cream cake that
was just offered to her. As Kendra’s mother, Marci should be familiar with Kendra’s
likes and dislikes, and it can be assumed that she chose the cake on the premise
that Kendra does like ice cream. The interrogative turn also functions as a means
for denying accountability for any mistake on Marci’s part for choosing a ‘wrong’
cake, as it displays that Kendra’s dislike of ice cream is news to her. Further, the in-
terrogative is produced in high pitch and with high-rising final intonation, which
indicates some degree of affect being involved in the turn (cf. Selting 1994) (due to
overlap it is unfortunately impossible to get an accurate pitch trace for the turn).
In short, the negative yes/no interrogative is used to challenge Kendra’s complaint
by calling into question her stance that she does not like ice cream; this position is
implicitly embedded in the complaint. Kendra’s response, I don’t `like ice ^crea=m
(line 7), asserts her stance explicitly. However, she does not provide an account
that would inform the others of what type of cake she does like, or why she does
not like ice cream. Wendy’s following turn, Do you like .. ^frozen `yogurt? (line 9),
can be seen to pursue such an account. It provides one candidate, ^frozen `yogurt,
for Kendra’s confirmation or disconfirmation. However, Kendra ignores Wendy’s
attempts, and moves on to the blowing out of the candles (lines 11–13). It is no-
table that Wendy returns to the issue after some intervening talk by reproducing
her question almost word-for-word, Do you like frozen yogurt cakes? As a result of
the talk that follows (see excerpt 8 below), Kendra almost apologizes for having
complained (lines 1–3); she also provides an account for not liking ice cream by
using reported speech (lines 18–19).
(8) (Appease the Monster SBCSAE 0013 <T: 00.19.35>)

1 KENDRA: I don’t want to hur=t you=,
2 I mean I like --
3 .. Ice [cream’s okay],
4 WENDY: [<HI Well,
5 MARCI: [I didn’t n- --
6 WENDY: we’re all] [2.. just HI>2] --
7 MARCI: You’ve] [2never2] told us what you like and don’t like,
8 [3dear,
9 KEVIN: [3I guess that would be=3],
10 MARCI: you just go off to your3] roo[4=m4].
11 KENDRA: [4I think4] [5I walk around all5]
the ti=me,
12 KEVIN: [4your4] [5fault,
13 then5].
14 KENDRA: saying,
15 WENDY: [(THROAT)]
16 KENDRA: [I hate] ice [2cream2].
17 WENDY: [2@@2]
18 KENDRA: I hate ice cream.
19 It makes me too cold.
In the previous example, the negative yes/no interrogative presented a positive

candidate understanding, i.e., the interrogative speaker displayed that she would
have supposed the recipient to like ice cream. In the next example, on the other
hand, the tag-question construction presents a candidate understanding of nega-
tive polarity. The conversation takes place between three participants. Harold and
Jamie are a couple, and Pete is their friend. Harold and Jamie have been telling
Pete at length about their neighbor and her children, all of whom they find to
be rather annoying. Previously, during the telling, Jamie has announced that the
neighbor is pregnant with another child. In the excerpt below, Harold returns to
this piece of news.
(9) (Lambada SBCSAE 0002 <T: 00:06:21>)

1 HAROLD: I `can’t t- --
2 `How do you --
3 She just ^looks `pregnant?
4 [`Now]?
5 JAMIE: [She’s] ^pregnant.
6 She’s ^totally `pregnant.
274 Tiina Keisanen
7 HAROLD: Oh.
8 JAMIE: .. It’s `not .. êating too `much,
9 she’s ^pregnant.
10 HAROLD: ^So=,
11 I `guess,
12 (1.2)
13 HAROLD: I `mean thi- this- thi- this just ^happened?
14 Or,
15 JAMIE: We’re gonna have `babies ^crying.
16 (0.4)
17 JAMIE: [].
18 HAROLD: [(GROAN)]
19 (1.2)
20 HAROLD: Well it’s no ^worse than her ^screaming at em,
21 ìs it?
22 (1.0)
23 PETE: `Yeah but `now you’ll have ^both.
24 (0.6)
25 JAMIE: `Yeah ^right.
26 (0.4)
27 Probably be like,
28 <VOX ^shut up you `ki- VOX>,
29 you `know,
30 XX?
31 Òh= ^Go=d.
On lines 1–4, Harold topicalizes the neighbor’s pregnancy that was first men-
tioned some time ago. He asks Jamie for the basis or the source of information for
her announcement that their annoying neighbor is pregnant again. Harold there-
by expresses doubt toward the accuracy of the news. Jamie responds not by giving
such an account, but by stating simply that She’s ^pregnant. She’s ^totally `preg-
nant (lines 5–6). Harold acknowledges this briefly with Oh (line 7). Jamie, in turn,
continues to attend to the inadequacy of her response by excluding one of the
most obvious counterarguments against pregnancy with .. It’s ^not .. êating too
`much, she’s ^pregnant (lines 8–9). Harold is still doubtful as he continues to work
toward creating an understanding of the state of affairs by presenting a request
for confirmation ^So=, I `guess, (1.2) I `mean thi- this- thi- this just ^happened?
(lines 10–13). However, Jamie does not respond to Harold’s inquiry. Instead, she
moves on to complain about the situation by mentioning one negative outcome
that the pregnancy will have on their lives, We’re gonna have `babies ^crying (line
15). After a pause, during which no one responds (line 16), Jamie adds, (line 17), thereby upgrading the complaint.
During the latter part of Jamie’s complaint Harold groans (line 18), which
shows recipiency and acknowledges that the complaint requires uptake. However,
the long pause (line 19) pre-monitors that the complaint will not get the align-
ing response it invites. Harold produces a tag-question challenge that targets the
basis of the complaint, Well it’s no ^worse than her ^screaming at em, ìs it? (lines
20–21). It presents a contrasting point of view on the issue by suggesting that it is
not only the noise that the children make, but also the noise that comes from their
mother screaming at them that should be taken into consideration when thinking
about the overall noisiness of the neighbors.
In addition to being a disaligning second-position turn, the tag question is
marked with turn-initial well to show disalignment or some discontinuity with the
prior turn (Pomerantz 1984b). Moreover, no worse sets up an opposition with the
prior turn, and the evaluative verb screaming explicates Harold’s negative stance
toward the noise the mother makes when communicating with her children.
As was the case in example (7), it can be noted that in the current example
Harold’s challenge is directed at the projected activity (complaining), and the ac-
tual content of the turn may be seen to include elements that show support or
the like. The challenge sets the complaint in a wider perspective by referring to
an earlier part of the conversation where Jamie herself portrayed the neighbor as
a person who yells and screams at her children. In other words, Harold uses the
tag-question turn to challenge the appropriateness of Jamie’s complaint by indi-
cating and/or reminding Jamie that the situation is already bad, and that it cannot
get much worse, no matter how much the neighbor’s new baby will cry. However,
instead of Jamie responding to Harold’s challenge, Pete takes a turn to counter the
challenge with `Yeah but `now you’ll have ^both (line 23). The turn indicates that
the two scenarios presented by Jamie and Harold are equally bad, and therefore
equally correct. In other words, Pete counters Harold’s challenge by challenging
its grounds. Jamie, in turn, seconds Pete with `Yeah ^right (line 25).
To sum up, similarly to Section 3.1, in the present section the challenging
tag questions and negative yes/no interrogatives likewise point toward some dis-
crepancy in information between the participants. In response to initiating ac-
tions such as complaints, the discrepancy is used to challenge the relevance or
the appropriateness of doing the action in the prior turn. As such, the negative
yes/no interrogatives and tag questions offer a disaligning response to the initiat-
ing action. It can be noted that even though both negative yes/no interrogatives
and the two types of tag questions are found to do a similar type of work in the
two sections above, the different constructions are not interchangeable. In excerpt
(2), for example, the challenging turn consists of a negative declarative followed
276 Tiina Keisanen
by a positive tag, and is used to challenge a claim (explicit or implied) or position

of positive polarity. On the other hand, the negative yes/no interrogatives in ex-
cerpts (5) and (7), for example, challenge a claim (explicit or implied) or stance of
negative polarity. However, the differences between the constructions are not well
studied, and other differences are also likely to exist. Future work on this topic is
necessary to elucidate the exact differences in more detail.
The examples in the sections above provide evidence that sometimes dis-
course participants can prioritize holding prior speakers accountable for their
claims and stances or positions the have taken, even implicitly, rather than align
with the projected course of action. So, even though the preference for agreement
(Sacks [1973]1987) limits the occurrence of disagreement in social interaction,
prior stances can be contested and challenged, as the interrogative constructions
discussed here demonstrate. Moreover, on the sequential level such challenges
can function as a vehicle with which to exert influence on the course of the un-
folding action.
Sequential positioning is the central factor in the interpretation of the nega-
tive yes/no interrogatives and tag questions discussed here as challenges. Howev-
er, in addition to sequential positioning, the placement of the interrogative turns
in the stream of talk also contributes to their interpretation. More specifically, the
smooth transition from a turn to a next with no gap and no overlap (Sacks et al.
1974) is regularly disturbed by the challenging negative yes/no interrogatives and
tag questions. In about half of the cases in the database, the challenges are placed
in overlap with prior talk (e.g., excerpts 3 and 4). Recipients are also observed to
rather frequently withdraw or delay their response to the challenging turns (e.g.,
excerpts 5 and 9). These are some further indications of the disaligning and dis-
preferred nature of the interrogatives discussed here.
The linguistic and prosodic turn-design features also contribute to the con-
struction of disalignment. First, the challenges are regularly designed to negate or
contradict a claim or position embedded in the prior turn, while the interrogative
design is used for requesting the recipient to account for this claimed discrepancy.
The turns may also include lexical markers, such as different (example 3), only
(example 4), really (example 2), and no worse (example 9), which help to identify
the problematic portion of the prior turn. In terms of linguistic practices of stan-
cetaking, these can be seen to function as restrictive stance markers in their local
context. On the other hand, the use of marked prosody, especially of high pitch,
can be used to add an affective stance to the challenging turns.
4. Conclusion
The speakers of negative yes/no interrogatives and tag questions discussed here
initiate a sequence that ostensibly concerns some piece of information. However,
due to the action of challenging carried out with the interrogatives, on the level
of social relations these sequences involve the issue of alignment between the dis-
course participants. What is notable is that the negotiation of alignment does not
surface in the interaction explicitly, but as I hope to have demonstrated, it is still
available to analysis through the examination of sequence organization, the dis-
tribution of silence and talk, as well as turn-design features of the turns in ques-
tion. In other words, the patterns of stancetaking discussed here expand on two
different, but interlocking levels. On the one hand, stancetaking can be conceptu-
alized as alignment or disalignment between discourse participants with respect
to the projected course of action or of the sequence. The interrogatives discussed
here provide the discourse participants with a resource for doing disalignment,
specifically to challenge the projected or ongoing course of action. On the other
hand, when the features of the aligning or disaligning actions are examined in
more detail, it is observed that actions may embody epistemic as well as affective
stance displays toward not only one’s own turn, but also toward others’ turns.
In the data here, the challenging actions are used to display doubt toward prior
claims or stances taken, in addition to which the prosodic turn-design features
can be used to add an affective component to the turn.
As demonstrated by Pomerantz (1984b) in her discussion on agreement in
assessment sequences, delay devices such as pauses and repair initiators are com-
monly found in disagreeing turns that are produced in response to turns that in-
vite agreement. Delays, repair initiators, and reformulations are found also in the
sequences discussed here. They indicate troubles in turn transition, and point to-
ward an interpretation that the recipients regard the interrogative turns as indica-
tive of disalignment. However, in telling sequences the troubles in uptake may also
lie in the fact that the interrogatives foreground something that was not originally
presented as contestable or challengeable information, such as, for example, the
information conveyed in statements of opinion often is. This may cause trouble
for the recipient in interpreting the relevance of an interrogative. That is, inter-
rogatives that are responsive to telling sequences can be seen as opportunistic in
assuming that a stance or a claim was provided for uptake when in fact it was not.
Such sequences draw attention to the fact that discourse participants can be held
accountable for anything that they produce in interaction, even implicitly.
The data discussed here provide evidence for stancetaking as an intersubjec-
tive achievement between discourse participants. By singling out some (explicitly
or implicitly expressed) claim or position in the prior turn, discourse participants
278 Tiina Keisanen
can call the interactional positioning implicated in the voicing of such claims or
positions into question. On occasion, then, taking a stance can be treated as an
accountable action.
Notes
1. This chapter is based on a section in my dissertation research (Keisanen 2006). Acta Uni-
versitatis Ouluensis is gratefully acknowledged for allowing the reprinting of the relevant ma-
terials. I would also like to thank the participants of the 10th Biennial Rice Linguistics Sympo-
sium held March 31 – April 3, 2004 at Rice University where the ideas for this paper were first
presented, and the Academy of Finland (project number 53671) for the funding of the research.
Moreover, John Du Bois, Robert Englebretson, Pentti Haddington, Elise Kärkkäinen, Maarit
Niemelä, Mirka Rauniomaa, Sandra Thompson and two anonymous reviewers have provided
constructive critique and insightful comments on the paper for which I am very grateful. All
the remaining inaccuracies are mine alone.
2. While outside factors to the ongoing conversation are not favored within conversation anal-
ysis, in this case the fact that the discourse participants are siblings seems to contribute to the
challenging nature of the tag question turn. In other words, as sisters the three women might be
expected to more or less know and remember each other’s prior teachers. Based on this knowl-
edge base that seems to be evoked and claimed as common (cf. Heritage 2002), the tag question
turn is produced from a strong epistemic position and thereby carries out a challenge (cf. Koshik
2003), rather than being used, for example, as an epistemically less strong (skeptical) newsmark-
er, which it might well be between less intimate participants. However, despite the label chosen
for the action (cf. Pomerantz 1990), the main argument here, as well as in the other cases, is that
the interrogative turn displays doubt towards the prior speaker’s claim or position.
3. Similar to example (2), the participants in this example are close relatives (cousins) based
on which Lenore can be seen to claim access to a shared knowledge base and on which she can
build an epistemically strong challenging turn. However, the high rising final pitch might be
used here to mitigate the degree of the hostility component that has been associated with chal-
lenging actions (Heritage 2002). One possible future direction for studies on these interrogative
forms might thus be to examine the import of the height and direction of the terminal pitch in
order to determine whether, for example, high rising final pitch can be used to make the turn
appear less hostile and/or challenging.
References
Beach, W. and Metzger, T. 1997. “Claiming insufficient knowledge.” Human Communication

Research 23 (4): 560–585.
Clift, R. 2000. “Stance-taking in reported speech.” Essex Research Reports in Linguistics 32. De-
partment of Language and Linguistics, University of Essex.
Drew, P. 1997. “‘Open’ class repair initiators in response to sequential sources of troubles in
conversation.” Journal of Pragmatics 28: 69–101.
Drew, P. 1998. “Complaints about transgressions and misconduct.” Research on Language and
Social Interaction 31 (3, 4): 295–325.
Pennsylvania.
Du Bois, J.W., Chafe, W., Meyer, C., Thompson, S.A. and Martey, N. 2003. Santa Barbara Cor-
pus of Spoken American English, CD, Part 2. Philadelphia: Linguistic Data Consortium,
University of Pennsylvania.
DVD, Part 3. Philadelphia: Linguistic Data Consortium, University of Pennsylvania.
Du Bois, J.W., Schuetze-Coburn, S., Cumming, S. and Paolino, D. 1993. “An outline of dis-
course transcription.” In Talking Data: Transcription and Coding in Discourse Research, J.A.
Edwards and M.D. Lampert (eds.), 45–87. Hillsdale, NJ: Lawrence Erlbaum.
Ford, C. 2004. “Contingency and units in interaction.” Discourse Studies 6 (1): 27–52.
Ford, C., Fox, B. and Thompson, S.A. 2003. “Social interaction and grammar.” In The New Psy-
chology of Language. Cognitive and Functional Approaches to Language Structure, M. To-
masello (ed.), 119–144. Mahwah, NJ: Lawrence Erlbaum.
Freese, J. and Maynard, D. 1998. “Prosodic features of bad news and good news in conversa-
tion.” Language in Society 27: 195–219.
Givón, T. 2001. Syntax: An Introduction, Vol. 2. Amsterdam: John Benjamins.
Goodwin, C. 2002. “Time in action.” Current Anthropology 43 (S4): S19–S35.
tures of Social Action. Studies in Conversation Analysis, M.J. Atkinson and J. Heritage (eds.),
299–345. Cambridge: Cambridge University Press.
Heritage, J. 1998. “Oh-prefaced responses of inquiry.” Language in Society 27: 291–334.
Heritage, J. 2002. “The limits of questioning: Negative interrogatives and hostile question con-
tent.” Journal of Pragmatics 34: 1427–1446.
Hopper, P.J. 2000. “Grammatical constructions and their discourse origin: Prototype or family
resemblance?” In Applied Cognitive Linguistics: Theory, Acquisition and Language Peda-
gogy, M. Pütz, S. Niemeier and R. Dirven (eds.), 109–129. Berlin: Mouton de Gryter.
Keevallik, L. 2003. From Interaction to Grammar. Estonian Finite Verb Forms in Conversation.
Ph.D. dissertation. Department of Finno-Ugric Languages, Uppsala University.
Keisanen, T. 2006. Patterns of Stance Taking: Negative Yes/No Interrogatives and Tag Questions
in American English Conversation. Acta Universitatis Ouluensis B 71. Oulu, Finland: Oulu
University Press.
Kim, K.-h. 1995. “Wh-clefts and left-dislocation in English conversation: Cases of topicaliza-
tion.” In Word Order in Discourse, P. Downing and M. Noonan (eds.), 247–296. Amster-
Koshik, I. 2003. “Wh-questions used as challenges.” Discourse Studies 5 (1): 51–77.
Koshik, I. 2005. Beyond rhetorical questions: Assertive questions in everyday interaction. Ams-
terdam: John Benjamins.
Pomerantz, A. 1984a. “Giving a source or basis: The practice in conversation of telling ‘How I
know’.” Journal of Pragmatics 8 (5, 6): 607–625.
280 Tiina Keisanen
Pomerantz, A. 1984b. “Agreeing and disagreeing with assessments: Some features of preferred/
dispreferred turn shapes.” In Structures of Social Action. Studies in Conversation Analysis,
Pomerantz, A. 1990. “Conversation analytic claims.” Communication Monographs 57 (3): 231–
235.
Raymond, G. 2003. “Grammar and social organization: Yes/no interrogatives and the structure
of responding.” American Sociological Review 68 (6): 939–967.
Sacks, H. [1973]1987. “On the preferences for agreement and contiguity in sequences in con-
versation.” In Talk and Social Organization, G. Button and J.R.E. Lee (eds.), 54–69. Clev-
edon: Multilingual Matters.
turn-taking for conversation.” Language 50 (4): 696–735.
Sadock, J.M. and Zwicky, A.M. 1985. “Speech act disctinctions in syntax.” In Language Typol-
ogy and Syntactic Description. Volume 1. Clause structure, T. Shopen (ed.), 155–196. Cam-
Schegloff, E.A. 1968. “Sequencing in conversational openings.” American Anthropologist 70 (6):
1075–1095.
Schegloff, E.A. 1988. “Goffman and the analysis of conversation.” In Erving Goffman. Exploring
the interaction order, P. Drew and A. Wootton (eds.), 89–135. Cambridge: Polity Press.
Schegloff, E.A. 1995. “Discourse as an interactional achievement III: The omnirelevance of ac-
tion.” Research on Language and Social Interaction 28 (3): 185–211.
Schegloff, E.A. [1995] 2007. Sequence Organization in Interaction: A primer in conversation
analysis I. Cambridge: Cambridge University Press.
Schegloff, E.A., Jefferson, G. and Sacks, H. 1977. “The preference for self-correction in the or-
ganization of repair in conversation.” Language 53 (2): 361–382.
Schegloff, E.A. and Sacks, H. 1973. “Opening up closings.” Semiotica 7 (4): 289–327.
Selting, M. 1994. “Emphatic speech style–with special focus on the prosodic signaling of height-
ened emotive involvement in conversation.” Journal of Pragmatics 22 (4): 375–408.
Thompson, S.A. 1998. “A discourse explanation for the cross-linguistic differences in the gram-
mar of interrogation and negation.” In Case, Typology, and Grammar, A. Siewierska and J.J.
Song (eds.), 309–341. Amsterdam: John Benjamins.
(from Du Bois et al. 1993)
Units
Intonation unit {carriage return}
Truncated word %
Final .
Continuing ,
Appeal ?
Speakers
Speech overlap (numbers inside
brackets index overlaps [2two words2]) [ ]

Primary accent ^
Secondary accent `
Unaccented
Lengthening =
Pause
Long and medium (N)
Short (brief break in speech rhythm) ..
Vocal noises
e.g., (TSK), (SNIFF), (YAWN), (DRINK)
Glottal stop %
Exhalation (Hx)
Inhalation (H)
Laughter during speech (1–5 words) @ (e.g. @two @words)
Laughter during speech (+6 words) @ (e.g. <@ six words @>)
Quality
Special voice quality <VOX VOX>
Forte: loud <F F>
Higher pitch level <HI HI>
Parenthetical prosody <PAR PAR>
Uncertain hearing <X X>
Positioning and alignment as activities
of stancetaking in news interviews
Pentti Haddington
1. Introduction1
This paper investigates two interactional stancetaking activities called positioning

and alignment in American and British news interviews. It represents a function-
al-interactional approach to studying stancetaking (Haddington 2004, 2005a),
which combines Du Bois’ (this volume) theory of stance with conversation ana-
lytic methods of investigating the sequential and interactional aspects of talk-in-
interaction and the grammar and grammatical practices for doing actions in in-
teraction (Schegloff 1996). Stancetaking is here understood as an intersubjective
activity in which interlocutors collaboratively construct and take stances about a
stance object, by engaging with and building upon stances taken by their co-par-
ticipants in the immediately prior talk (Du Bois this volume; Haddington 2005a).
When interlocutors take stances, they place themselves with regard to others and
stance objects, but also engage and align with previous interlocutors and their
stances (Du Bois this volume).
I concentrate on describing the linguistic, sequential, and turn organizational
features of positioning and alignment. Positioning or setting up a position is an
activity in which the interviewer designs a question so that answering it poses
difficult problems for the interviewee. Consequently, it is also a forward-type in-
tersubjective activity, because it establishes and creates an interactional context
for the interviewee. A positioning question encodes or evokes a preferred stance,
presumption, or a presupposition that puts the interviewee in a quandary relative
to her institutional background, a publicly taken prior stance, audience expecta-
tions, and broader sociocultural beliefs and so on. Although the propositional
context of the co-participants’ stance is not of interest per se, it is worth consid-
ering the contextual background and the indexical properties of language, and
284 Pentti Haddington
how they together with other practices invoke particular situated interpretations
(Haddington 2004). The starting point for looking at positioning is the under-
standing of questioning as a complex activity in which several – and not single –
turn constructional units (TCUs) affect the understanding of the question and the
trajectory of the ensuing talk (Heritage and Roth 1995).2 Clayman and Heritage
(2002: 209) refer to the possibility that combinations of practices in questions may
strongly favor particular answers over others, but do not discuss this possibility
in their examples.
Nevertheless, previous conversation analytic (CA) work on news interview
interaction has investigated how interviewers use various practices and actions in
order to question their guests in different ways. This body of work has described
how interviewers on the one hand design their questions as neutralistic (Clayman
1988; Clayman and Heritage 2002: 126–131; Heritage and Greatbatch 1991), and
on the other hand, how they also adopt adversarial stances and exert pressure
on their respondents by incorporating third-party statements, particular topical
agendas, presuppositions, and accusations in their questions (Berg 2003, 2001;
Clayman and Heritage 2002; Heritage 2003; Heritage and Roth 1995). However,
many of these studies have only described the ways in which these practices and
actions are accomplished within a TCU, and have thus given little attention to the
design of entire questioning turns in news interviews (cf. question delivery struc-
tures, Heritage and Roth 1995) and the recurrent linguistic practices in them.
In fact, Schegloff (1996) suggests that a linguistic vantage point can be used for
looking at how speaker turns are composed of individual TCUs, how TCUs are
sequentially positioned in their turns, how they are related to each other, and how
all this affects the understanding of the turns (Schegloff 1996: 64). Similarly, Ford
(2002, 2004) shows that it is fruitful to consider how recurrent action combina-
tions and linguistic formats in turns are utilized for constructing co-participation
in interaction. Schegloff ’s (1996) and Ford’s (2002, 2004) work provide a use-
ful starting point for considering how the interlocutors’ stancetaking is organized
across TCUs, transition relevance places (TRPs) and speaker turns.3 It also en-
ables us to consider how the combinations of TCUs, and the actions within those
TCUs, contribute to the interlocutors’ stancetaking. This approach, which goes
beyond the examination of individual TCUs, is particularly relevant for the inves-
tigation of news-interview interaction, because talk in news interviews is largely
composed of multi-unit turns (cf. Heritage and Roth 1995; ten Have 2001: 6).
By recognizing the practices by which interviewers set up positions for their
guests, it is also possible to understand the underlying motivations behind some
instances of interviewee resistance. Although the examples of the interviewees’
answers in this paper could be seen to be evasive, it is also fruitful to consider the
ways in which they engage and align with the question. The notion of alignment is
Stancetaking in news interviews 285
here understood somewhat differently from its use in CA. In CA, an aligning ac-
tion is understood to be an appropriate or preferred next action, which fulfills the
expectations raised by the previous action. For example, an aligning action to a
request is an acceptance, and a disaligning action is a refusal (Heritage 1984: 269).
The notion of alignment here is also not synonymous with agreement. Rather, it
aims to explicate the range of possible types of convergent and divergent positions
that interactants can take relative to each other, i.e., “how I put my stance in rela-
tion to your stance” (Du Bois, p.c.). Interlocutors use language for aligning with
each other and therefore, alignment, as it is outlined by Du Bois (this volume) and
adopted here, is very much a linguistic process in which interactants use morpho-
syntax, lexis, and prosody to construct their stances.
One central element of linguistic alignment, as Du Bois (2001) further claims,
is that speakers frequently use and recycle the linguistic elements that their co-
participants have used. By recycling linguistic forms, speakers bring two utter-
ances into close relation to one another. When this happens, these forms engage.
This paradigmatic connection between two utterances can generate new local
meanings, even ones that at first sight seem disparate. Thus the connections be-
tween the “formal” structures of language can naturally affect the way in which
individual stances are understood in the interactional context (Du Bois 2001).
And as Du Bois (2001) shows, interlocutors use linguistic forms dialogically in
order to negotiate their stances and create new meanings. It is noteworthy that
by recycling elements, speakers do not necessarily aim to display agreement (Du
Bois 2001). Rather, by recycling linguistic forms, interactants can frequently dis-
play subtle differences between their stances.
In news interviews, the interviewee’s alignment activity can be seen as a back-
ward-type intersubjective activity, and a contingent achievement by which she
responds to the position in the question. Although conversation analysis has de-
scribed such “evasive” interviewee conduct, interviewer responses to such eva-
sive maneuvers (Clayman 2001: 238–298; Clayman and Heritage 2002), change of
topical agendas, and question reformulations (Clayman 1993; Greatbatch 1986),
it has largely looked at them separate from the question. In comparison, this pa-
per focuses on some aspects of the interviewer’s turn that act as an impetus for an
indirect answer, and some of the interviewees’ ways of responding to and answer-
ing difficult questions.4 One of the most important claims in this paper is indeed
that it is the design of the question and the stance it incorporates that strongly
affects the interviewee’s subsequent turn and the stance therein. As we will see,
interviewees use various types of turn-design features, practices, and actions to
respond to difficult questions and to intersubjectively engage with them.
In sum, this paper supplements previous work by examining positioning and
alignment from the following vantage points:
1. It looks at how the combinations of practices, actions, action combinations,

and particular turn-organizational features are used as resources and contrib-
ute to stancetaking in news interviews.
2. It describes a patterned and reusable linguistic practice that interviewees use
to manage their stancetaking in news interviews. More specifically, inter-
viewees frequently use a stance marker (with cognitive or communication
verbs) and a recycled element of the interviewer’s question, to align with a
proposition or presupposition incorporated in the question. Moreover, the
interviewee’s turn in these answers is organized into a negative-positive struc-
ture, which helps the interviewee not only align with the question but also
construct a stance of her own.
3. It shows how the co-participants intersubjectively engage in stancetaking.
This paper supports prior findings (Du Bois this volume; Kärkkäinen 2003b) that
by taking a stance an interlocutor (in this case the interviewee) displays an under-
standing of a prior stance (in the interviewer’s question). Thus, the findings in this
paper support the claim that stancetaking is a contingent achievement occasioned
by prior stances. Finally, I conclude that the positioning and aligning activities
show that stancetaking in news interviews is not as emergent or performed as it is
in everyday talk, but rather that the interviewers come to the interview situation
with certain predetermined questions and agendas in mind and the interview-
ees – although they naturally construct their stances vis-à-vis the interviewer and
the question – design their answers based on their own objectives and attitudes.
The database used in this paper was collected between October 1999 and
March 2004. The corpus contains approximately 20 hours of news interviews from
both the United States and Britain. The data fragments are transcribed in the Dis-
course Transcription (DT) style in which one line represents one intonation unit
(Du Bois et al. 1993). The transcription conventions are given in the Appendix.
2. Practices of positioning and alignment in news interviews
In the following sections, I discuss how interviewers set up difficult positions for the
interviewees by asking hostile questions, evoking and encoding preferred stances,
and by incorporating presuppositions in the questioning turns. I also identify some
recurrent practices by which interviewees align with these positions.
2.1 Positioning by hostile questions and interviewees’ practices

of aligning with them
In example (1) below, which is from BBC2’s Newsnight, Jeremy Paxman inter-
views Alistair Darling, the Secretary of State for the Department for Transport. It
presents an interesting example of how the interviewer positions the interviewee
by designing a somewhat hostile question that the interviewee cannot answer di-
rectly. The question is about whether the British government should intervene
in the tendency of falling flight prices.5 Nevertheless, the interviewee does not
bluntly evade the question but manages to align and engage with it.
(1) BBC2, Newsnight, Dec 17, 2003: Cheap air travel
IR: Jeremy Paxman, IE: Alistair Darling (033 / 1 / 0:34)
1 →2 IR: Well earlier I spoke to the transport secretary Alistair Darling,
2 (TSK)(H) I kicked off by asking him,
3 →1 .. if he thought air travel (H) was too cheap.
4 ((CUT TO ALISTAIR DARLING))
5 ...(1.4)
6 IE: I think that aviation,
7 ought to=,
8 meet the= `costs of the environmental damage it ^%causes.
9 Although .. (GLOTTAL) the ^difficulty there is that,
10 taxation of aviation fuel for example,
11 (H) is ^dealt with by ìnternational ^treaty,
12 and we ^can’t do sen- something,
13 (H) %uh [u]nilaterally.
14 IR: [S-] --
15 IE: (H) But in ^relation to=,
16 aviation costs in the last few years,
17 the reason it’s come ^down,
18 (H) is because,
19 the industry has ^dramatically cut its ^costs,
20 and passed that on to the passengers.
21 →3 IR: (0) Do you think it’s too cheap.
22 IE: .. (H)(TSK) No I do- I- --
23 <A>Now I would</A> --
24 →4 I wouldn’t argue that it’s too cheap,
25 (H) I ^do think though,
26 .. <MRC>it ought to meet the costs of </MRC>the environmental
27 damage,
28 that it causes.
The interviewer’s first question in lines 2–3 is a reported yes/no question. It is not
possible to know whether the wording of the actual question was the same. If it
was not, this example provides a rare instance of how editing is used for changing
the meaning of the question in news interviews and naturally has implications
for the following analysis.6 Nevertheless, the design of the subsequent interaction
suggests that the original question cannot have differed greatly from the reported
question. Especially the use of the it-pronoun in the interviewer’s question in line
21 seems to refer back to the (unanswered) previous question.
The reported question is designed so that both of the projected alternative an-
swers (an agreement or disagreement) are potentially harmful for the interviewee.
As a transport secretary, the interviewee has to consider the interests of the airline
companies, which of course makes giving a negative answer problematic. How-
ever, as an elected politician, answering yes would obviously be counter to the
general public, which prefers cheap flight prices. Thus, in terms of positioning,
the interviewer simultaneously uses the yes/no question (arrow 1) together with
the interviewee’s identity (arrow 2) in order to put the interviewee in a problem-
atic situation.
At first, the interviewee chooses to evade the question altogether (lines 6–13)
by raising the issue of the environmental damages that are caused by aviation
(cf. Clayman 2001; Clayman and Heritage 2002: 238–298). After this, the inter-
viewee moves (lines 15–20) toward the original topical agenda in the question.
However, he only reports some reasons why flight prices have come down and, in
effect, does not answer the question. Then (in line 21) the interviewer orients to
the interviewee’s dodging action and repeats the question. The repeated question
therefore resurrects the position setup in the first question with the additional
element of holding the interviewee accountable for not answering the question
the first time.
Then in lines 22–23, the interviewee begins to align with the question. Judg-
ing from the truncations and the hesitation in the beginning of the interviewee’s
answer, he at first has problems in formulating his response to the interviewer’s
hostile repeated question. He does not finish the already started utterance No I
do- I- -- that projects the utterance No I don’t think and goes on to modify the
way he frames his stance. He does this by producing a self-repair and begins to
produce an alternative modal auxiliary wouldn’t. Consequently (arrow 4), instead
of recycling the verb think, the interviewee uses the communication verb argue
(Biber et al. 2003: 316) in line 24. It could be that the change of the modal auxil-
iary is closely connected to the change of the verb as well. It is likely that the inter-
viewee cut off a denial that was going to be produced as the stance marker I don’t
think, but then changed it to a different stance marker I wouldn’t argue. By fram-
ing his stance in this way, the interviewee refrains from producing a direct denial
to the question (which I don’t think it’s too cheap would have done), and rather
claims that he would not engage in an argument about the issue. In other words,
a denial produced by the utterance I don’t think it’s too cheap would communicate
a more personalized and a more certain stance than the denial I wouldn’t argue
that it’s too cheap seems to express. Therefore, the self-initiated self-repair and the
new stance marker wouldn’t argue seem to reflect the interviewer’s quandary with
respect to the position he has been put in; he must word his stance carefully in
order to take into account the airline companies’ and the public’s possibly differ-
ing opinions.
Stance markers with a first-person subject pronoun combined with a cogni-
tive verb (e.g., think, know, guess, and accept) or a communication verb (e.g., argue,
claim) are not only frequent in news interviews, but also play a significant role in
interviewee alignment. As many studies on different languages have shown, the
collocation between a first-person subject and a cognitive verb is strong in every-
day talk, and these expressions tend to precede complement clauses (Kärkkäinen
2003a, this volume; Karlsson 2003; Rauniomaa this volume; Scheibman 2001).
However, it would seem that speakers are not specifically referring to themselves
when using these markers. Rather, these markers are used as epistemic/eviden-
tial/evaluative fragments in contexts where a stance is going to be produced by the
speaker (Thompson 2002). This view is also supported by Simon-Vandenbergen
(2000), who shows that in news interviews, the stance marker I don’t think is used
for conveying and framing the speaker’s subsequent opinion, and by Schegloff
(1996: 61–62) who points out that in everyday talk, I don’t know seems to project
more talk to come. As the above example and the examples below show, these
types of stance markers occur in a specific context in news interviews: they pre-
cede a complement clause that is composed of language recycled from the inter-
viewer’s question turn. In the above example, the interviewee recycles the phrase
it’s too cheap (arrow 4) from the interviewer’s turn. In Diagraph 1 below, we can
see how the interviewee uses the stance marker I wouldn’t argue and the recycled
phrase in order to align with the question.7
Diagraph 1 (from example 1): Cheap air travel

21 IR: you {do} think it ’s too cheap
22 IE: No I do- I-
23 Now I would --
24 I wouldn’t argue that it ’s too cheap
The above diagraph clearly demonstrates Greatbatch’s (1986) and Clayman’s (1993;
2001) findings that interviewees frequently recycle a unit from the question in
order to shift the topical agenda or to evade the question altogether. Thus, this
practice has a clear function. By repeating and incorporating linguistic elements
from the question the interviewees show that they are attending to the question
and on the surface level answering it (Clayman and Heritage 2002: 247; Heritage
and Roth 1995). However, an alternative reading that considers the impact of the
question on the answer is perhaps more accurate in example (1). In other words,
answers that seem to just shift the topical agenda or evade the question altogether
often in fact intersubjectively engage and carefully align with the difficult position
the question sets up. Therefore, it is the way in which the stance is formulated
in the question that affects the design of the answer. The engagement and align-
ment of the answer with the question can be perceived in the linguistic pattern in
example (1) and Diagraph 1 (stance marker + recycled unit from the question).
This pattern is frequently used by interviewees for aligning with a difficult posi-
tion posed by the interviewer’s question. By using this pattern, the interviewee
in the above example locates the trouble source, and at the same time displays
fine-tuned and careful alignment with the position in the question and avoids the
pitfall that the interviewer laid out before him. And even though the interviewee’s
turn, in example (1) above, is constructed so that it displays a more explicit en-
gagement with the question than his previous answer (in lines 6–20), he basically
gives the same answer as he gave before the interviewer’s repeated question. In
other words, by carefully designing the answer so that it engages with the difficult
question, the interviewee rather than just evading the question, displays careful
alignment with the question and, in fact, answers it.
2.2 Positioning by incorporating preferred stances in questions

and practices of aligning with them
In the following, I show how the interviewer sets up a position for the interviewee
by designing the turn so that it evokes a preferred stance. The notion of prefer-
ence is one of the central analytic concepts in CA (cf. Hutchby and Wooffitt 1998;
Sacks [1973] 1987; Sacks and Schegloff 1979). It refers to the idea that in interac-
tion particular actions invite (i.e., prefer) particular types of responses (Hutchby
and Wooffitt 1998: 44). Preference does not refer to the co-participants’ internal
or subjective preferences, but rather to the structural features of how a turn or an
action is produced in its sequential context. The clearest evidence for preference
organization can be seen in adjacency pairs. For example, an invitation prefers an
acceptance; assessments prefer agreeing second assessments, and so on. Second
pair parts in adjacency pairs that are structured as preferred tend to be unmarked,
whereas dispreferred turns often incorporate different types of dispreference mark-
ers (cf. Hutchby and Wooffitt 1998: 44; Sacks [1973] 1987). Preference is also often
signaled with particular use of syntax. This is particularly clear in news interviews
in which negatively formulated questions are “routinely treated as embodying a
very strong preference for a yes answer” (Clayman and Heritage 2002: 209).
The question of whether a stance incorporated in the first turn can evoke a
preferred next stance has not yet received much attention. Since stancetaking is
not an action but a larger activity (Haddington 2004) the whole idea of prefer-
ence as it relates to stancetaking requires reconsideration. In other words, a stance
preferring a particular subsequent stance is a different issue from one action pre-
ferring a certain type of next action. Sacks ([1973] 1987: 58) rightly claimed that
features of preference organization can be examined without knowledge of the
participants, the topic, and where a sequence takes place. However, preference as
it relates to stancetaking benefits from considering at least some of the extra-tex-
tual and participant-specific aspects in the analyzed segment, as is shown below.
In the following example, the interviewer asks a question that is framed as a
third-party stance. This third-party stance and the design of the question evoke a
preferred stance in the response that the interviewee clearly orients to and aligns
with. This example comes from CNN’s Larry King Live. The third-party stance
in the question preface is based upon recent polls on whether the United States
should attack bin Laden in Afghanistan. After this, the interviewer, Larry King,
formulates a yes/no question to the interviewee Brian Jenkins.
(2) CNN, Larry King Live, Sep 12, 2001: Bomb him out
IR: Larry King, IE: Brian Jenkins (001 / 5 / 3:49)
1 IR: ...(0.6)(TSK) ^Brian,
2 could you=,
3 →1 ... well I guess the `public,
4 →1 would look at this `simply,
5 Could you ^bomb him out.
6 IE: .. (TSK)(H)U=h,
7 I d- --
8 →2 I don’t know that you could ^bo=mb him out,
9 I mean,
10 I I ^think that,
11 uh,
12 .. the ^magnitude of this operation,
13 is going to `call for,
14 a ^qualitatively different `response,
15 than we have seen in the ^past `two=,
16 two acts of ^terrorism.
The TCU, which the interviewer starts in line 2, projects a yes/no question. How-
ever, he abandons it and instead produces a TCU (lines 3–4) that frames the
newly-produced question TCU (in line 5) as a stance by a third-party. It has been
shown that interviewers use third-party statements in order to raise controversial
topics, but since these stances are attributed to third parties, the interviewers still
maintain a neutralistic stance toward the topical agenda and the guest (Clayman
1988; Heritage 2003; Heritage and Greatbatch 1991).8 In the above example, the
third-party stance is placed sequentially before the question (at arrow 1), and
therefore, presents the question not so much as the interviewer’s question, but as a
stance that the majority of the general public share. The third-party is mentioned
in the complement clause in lines 3–4. The complement clause is preceded by the
epistemic stance marker I guess in line 3. Kärkkäinen (this volume) claims that I
guess often frames actions in first position (e.g., questions) that invite second-pair
parts, and that I guess arises from some information or evidence that has been
established in prior discourse or is based on a private reasoning process of the
speaker. Indeed, not only does the marker here occur in first position (a question-
ing turn), but the cut-off IU and the use of the third-party stance also mark the
interviewer’s sudden shift from asking a simple information-seeking question to
asking a question on behalf of the audience, of which according to the polls at the
time 94 percent favored a military response. The sudden transformation of the
question’s design and the invoked third-party stance change the way in which the
question can be understood, and thus have important interactional consequences
for stancetaking. In other words, in addition to the fact that the interviewee has
to answer the question, he now also has to orient to the third-party stance given
in the question preface. This is because although the question TCU Could you
^bomb him out. in line 5 is a yes/no question and makes relevant either an af-
firmative or a negative answer, in this case, a subsequent negative answer would
mean that the interviewee disagrees with a large portion of the public and the
TV-viewers, a position one does not seek willingly. In other words, not only does
the question favor an agreeing answer (Sacks [1973] 1987), but the question TCU
together with this particular third-party stance (i.e., the combination of the two
actions) also strengthens the preference conditions for an answer that agrees with
the question.
The way in which the interviewer designs the questioning turn as a whole to
favor a particular answer would seem to set up a difficult position for the inter-
viewee. The evidence for this can be found in the beginning of the interviewee’s
turn. He goes through some trouble in formulating an answer to the question,
which is indicated by the pause before the interviewee’s answer, the hesitation
marker U=h (line 6), and the self-repair (line 7). These dispreference markers and
faltering talk are indicative of a stance that is divergent from the public stance
reported in the question. After this, the interviewee answers the question by ut-
tering I don’t know that you could ^bomb him out (arrow 2). As Diagraph 2 below
shows, after the stance marker, the interviewee again recycles an element from the
question, which establishes a connection between the question and the answer.
Diagraph 2 (from example 2): Bomb him out

5 IR: {you} could bomb him out
7 IE: I do- --
8 I do n’t know that you could bomb him out
The interviewees’ formulation of the TCU in line 8 is interesting, because the

recycled part of the TCU is a finite dependent that-clause in which the verb is in
the subjunctive mood.9 The use of subjunctives is very rare in spoken English,
but they are sometimes used to express unreal or hypothetical meanings (Biber
et al. 2003: 261). In comparison to a conditional clause (I don’t know if you could
bomb him out), which in this context would display epistemic uncertainty or in-
sufficient knowledge about the issue (Haddington 2005b), the interviewee now
orients to the preferred stance by a subtle expression of doubt about the public’s
view. In light of this, the use of the that-clause and the subjunctive contributes to
the aligning activity in an interesting way. It treats the third-party stance as doubt-
ful or even hypothetical, even though the reported third-party stance was actu-
ally based on topical polls. After this, the interviewee produces an explanation by
introducing alternative actions to “bombing.” In sum, the interviewee orients to
the preferred stance and shapes the answer so that it disagrees with it as little as
possible. In front of the audience (the public referred to in the question), this is
much more convenient for the interviewee than an explicit disagreement or an
evasive answer would be.
Example (3) below comes from CNN’s Crossfire.10 It was recorded in Decem-
ber 2002, a couple of months after the terrorist attacks on New York and Wash-
ington. The interviewer is Paul Begala, who served in the Clinton administration,
and the interviewee is Frank Gaffney, who is the former Assistant Secretary of
Defense and an expert in foreign and defense policy. The question concerns the
so-called “bin Laden tapes” and whether these tapes include covert messages for
inactive terrorists to take action and, therefore, should not be shown on televi-
sion. In this example, the interviewer positions the interviewee by invoking a
third-party stance, subtly incorporating his own stance, and designing the ques-
tion to contain a preferred answer. The question also shows how the interviewee
carefully aligns with this rather complex positioning activity.
(3) CNN, Crossfire, Dec 27, 2001: The new bin Laden tape
IR: Paul Begala, IE: Frank Gaffney (003 / 1 / 1:13)
1 IR: .. Uh the ^new bin Laden tape.
2 (H) ... that’s,
3 aired by Al-Jazeera ^today?
4 ...(H)(TSK) When,
5 .. he ^began sending these tapes out,
6 →1 .. the President’s national security adviser,
7 →1 ^told the `networks.
8 →1 ...(0.7) They ^shouldn’t run these,
9 →1 because `she ^feared,
10 →1 ... ^first that he would whip up --
11 →1 .. uhv,
12 →1 uhm,
13 →1 .. anti-American `views,
14 →1 but then ^second,
15 →1 and probably more .. òminously,
16 →1 (H) that there were <MRC>secret coded messages</MRC>,
17 →1 ^potentially in these tapes.
18 →2 ... <A>Now they’ve had some of these tapes for eight weeks.
19 →2 We have the best cryptographers in the world</A>.
20 →3 (0) Is there <MRC>slightest shred of evidence,
21 →3 that she was right</MRC>?
22 →4 IE: ...(1.0) (H) (TSK) Well ^nothing,
23 →4 .. has been blown up,
24 →4 ... so 'fa=r.
25 →4 .. So,
26 →4 ... you might deduce from that,
27 `no.
28 →5 ... I don’t think this is a matter of ^cryptography,
29 I think this is a ^question of whether,
30 <MRC>people have gotten instructions</MRC>,
31 ...(0.7) (H) who are ^here in this country,
32 or perhaps êlsewhere in the world,
33 ... (H) and will be prepared to ôperate on them.
The above interviewer’s turn is again composed of two main parts: the question
preface (in lines 1–19), which sets up the topical agenda and gives background in-
formation to the question, and the actual question (lines 20–21). At the beginning
of his turn, the interviewer introduces the stance object “the new bin Laden tape”
in lines 1–3, after which, in lines 6–17, he goes on to report a third-party stance,
i.e., what Condoleezza Rice, the President’s National Security Adviser, has said
about the stance object (that the networks should not run these tapes, and that
there are potentially secret coded messages in them). After this, the interviewer
concludes the turn by formulating an interrogative in lines 20–21. Even though
the question is not hostile or adversarial, the combination of the several practices
in the question set up a position for the interviewee.
First, similarly to example (2), in example (3), the interviewer’s action of
bringing up a stance by a third party (at arrows 1) already builds up a connection
between the third party and the interviewee. In December 2001, when this inter-
view was recorded, politicians across the political field (as well as the majority of
the American public and the TV networks) were almost unanimous in their sup-
port of the President and his administration. The connection between the third
party and the interviewee is made explicit in the question (arrows 3) in which the
interviewer requests the interviewee to respond and take a stance in relation to
the third-party stance (note the use of the pronoun she).
Second, the interviewer makes a footing shift (Goffman 1981) and produces
an assessment, which presupposes that the issue is about cryptography (We have
the best cryptographers in the world.) (arrows 2). These two TCUs are produced in
distinctly rapid speech, which suggests that he slips them in between the question
preface and the question proper. Nevertheless, the fact that the footing shift is
sequentially positioned before the actual question frames the question in a par-
ticular light. Even though the interviewer in these TCUs does not (and cannot)
express explicit disagreement with the third party, he frames the question by pro-
viding additional background information that undermines the reliability and the
correctness of the third-party stance. Furthermore, as we will see later, when the
answer is considered, these two TCUs indeed are contextually relevant and affect
the way in which the following yes/no interrogative is understood.
Finally, the question that ends the interviewer’s turn in lines 20–21 contains a
preference. It is designed as a yes/no-type interrogative which narrows down the
possible relevant answers to an affirmative yes or a negative no. The question also
contains the adjective slight in the superlative form and the noun shred, which
together invoke the idea of smallness and insignificance. These words could be
perceived as negative polarity items (Horn 1989), which usually embody a prefer-
ence for a no answer (Clayman and Heritage 2002: 211–212). The preference for
the negative answer is also strengthened with the superlative construction. The
negative polarity items and the superlative construction together suggest that in
spite of all the intelligence resources, Condoleezza Rice does not have any evi-
dence to support her argument that the interviewer had reported in lines 6–17.
This is further emphasized by the interviewer’s voice quality: each word in this
unit is distinct and emphasized (marcato voice quality) and the interviewer’s pitch
is considerably higher than in the other parts of his turn. In sum, the third-party
statement, the sequential proximity of the footing shift and question TCU, and
the linguistic design of the question together contribute to the positioning activity
that the turn is doing. The interviewee is basically asked to display his disagree-
ment with the President’s National Security Adviser, which at that historical mo-
ment could have displayed the interviewee in an unfavorable light.
In spite of the difficult position, the interviewee does not evade the question,
but rather addresses and responds to both parts of it: on the one hand, the issue
regarding Condoleezza Rice (lines 22–27) and, on the other hand, the cryptogra-
phy/secret coded messages-issue (lines 28–33). In addition to this, the interview-
ee orients to the question’s problematic preference structure and tacks through it
by designing the answering turn in a very careful manner.
As we saw above, the question prefers a negative answer. The interviewee com-
plies with this by eventually answering no in line 27. However, the interviewee re-
sorts to dispreference markers – the one-second pause and the adverb well – that
display the trouble he has in responding to the question and the stance therein.
In addition to this, before producing the no-answer, the interviewee carefully de-
signs the first two TCUs (at arrows 4) so that he avoids encoding them as personal
stances and rather relies on evidential information (nothing, .. has been blown
up, ... so ‘fa=r.) and induction (you might deduce from that,) (cf. Chafe 1986). He
further distances himself from the stance he takes by using the generic pronoun
you and the modal verb might (cf. can). Thus, even though the interviewee pro-
vides the preferred answer no, the two preceding TCUs frame the answer as a
depersonalized stance. Only after this does the interviewee continue to respond
to other parts of the question (cf. fn 4 in Clayman and Heritage 2002: 106–107).
He does this by denying the presupposition about the relevance of cryptography
(arrow 5), and then providing his view of what the issue is really about. He does
this again by first producing the stance marker I don’t think, which is followed by
a recycled lexical item in a slightly modified form (cryptography) from the ques-
tion in line 28.
Diagraph 3 (from example 3): The new bin Laden tape (question preface)
19 IR: We have the best cryptographers {}
28 IE: I don’t think this is a matter of cryptography
The interviewee produces the denial by first locating a trouble source in the ques-
tion preface and by repeating it, the noun cryptography, in his turn. By doing so,
he denies the presupposition in the question and its relevance in relation to the
issue at hand. This precedes and strongly projects a resolution or an account to the
denied issue (Ford 2002; Haddington 2005b), which is then provided immediately
after the denial. In sum, the interviewee not only orients to the preferred stance
evoked in the question, but also uses the interactional space he has been allocated
to design and present the stance in his own terms. In other words, the above prac-
tices, which the interviewee relies on, imply that he orients to the preferred stance
in the question and agrees with it (although minimally). However, at the same
time, these practices turn out to be handy for expressing an alternative stance.
In sum, the two examples in this section have shown different ways in which
interviewers incorporate preferred stances in the questions, and thereby put their
guests in difficult positions to answer them. In spite of this, the interviewees do
not just avoid answering the questions, but in fact intersubjectively align with the
position set up in the question. In the following section, I concentrate in greater
detail on how interviewers use presuppositions for setting up positions for the
interviewees and how interviewees deny them.
2.3 Using presuppositions for positioning and linguistic practices

for aligning with them
As Clayman and Heritage (2002: 127, 203–208) note, and as we saw in example

(3), interviewers can and often do embody presuppositions in the questions. Ac-
cording to Levinson (1983: 167), presuppositions are pragmatic inferences that
are based on linguistic structure and are sensitive to contextual factors. Here a
presupposition is understood to be a background assumption, which, on the one
hand, is used in the interviewer’s question for rendering the question sensible
and understandable for the interviewee and the overhearing audience, and on the
other hand, for setting up a position for the interviewee to take a stance. Clayman
and Heritage (2002) claim that in news interviews presuppositions differ in terms
of their “embeddedness” in the question. Some presuppositions are closer to the
surface than others, and some are easier to deny than others. Although presuppo-
sitions are not “explicit” stances, but rather compositions of “given” information,
“built-in” elements, and “self-evident” assumptions on which speaker stances are
built, the following examples show that the subsequent speaker can orient to a
presupposition. In other words, a presupposition, which is incorporated in an ut-
terance, can affect the next utterance and the stance in that utterance.
In the following, I briefly go through examples in which presuppositions are
used for setting up a position for the interviewee. It is worth noting that in the
following questions, presuppositions are not the only practice used for setting up
a position for the interviewee, rather the interviewers actually use a combination
of various practices to construct a problematic position for the guest. After that,
I consider the resources with which the interviewees deal with the presupposi-
tions. As in the examples above, the interviewees use the same linguistic practice
(stance marker + linguistic recycling) for aligning with the question. All in all,
my aim is to show that even though the answers can be claimed to be somewhat
evasive, the answers intersubjectively orient to the position set up in the question
both linguistically and in terms of turn design.
Example (4) below comes from Larry King Live. Here, the interviewer asks a
question from Christopher Whitcomb, a special agent for the FBI. The question
concerns Osama bin Laden, the alleged leader of the Al-Qaida terrorist organiza-
tion, and whether bin Laden should be killed or taken captive and then brought
to justice in the United States.
(4) CNN, Larry King Live, Sep 12, 2001: Rid them from this planet
IR: Larry King, IE: Christopher Whitcomb (001 / 5 / 1:24)
1 IR: Christopher,
2 if it ^gets to the `point,
3 →1 as we have ^promised that,
4 wl- --
5 →1 `we will ^take them out,
6 →1 we will=,
7 u=h,
8 <MRC>rid them from this planet</MRC>,
9 in a sense,
10 →1 We are at ^war with them,
11 →2 ^How `do you `do that Christopher,
12 →2 ^How do you ... ^take out a,
13 →2 a ^bin Laden or `someone like a bin Laden,
14 who’s hidden,
15 who `moves arou=nd,
16 and who ^may be under ^cover of the `country he lives in.
17 →3 IE: ...(0.8) (TSK)(H) Well Ì don’t know about ^taking him out Larry,
18 But I think we have a ^lot of òptions,
19 And one of them òbviously is ^military.
20 (H) The ôther=,
21 because this is a ^law enforcement òperation,
22 and because `we ^have to=,
23 (H) uh `go with the ^justice system,
24 and we want to bring ùltimately these people to ^justice,
25 (H) <L>Uh we’ve ^done,
26 in the ^pa=st</L>,
27 (H) òperations which are referred to as ^renditions,

28 where we go into òther ^countries,
29 (H) with the ^cooperation of those countries of course,
30 and bring people ^back to the United States.
The question contains two distinct units: a question preface in lines 2–10, which
provides a contextual background for the question, and the actual question in
lines 11–16. The question contains two presuppositions, which are not hostile
(Clayman and Heritage 2002; Heritage 2003, 2002) but nevertheless set up a posi-
tion for the interviewee. First of all, the question preface contains stances that are
attributed to we (arrows 1). This pronoun, which can also be called a shifter (cf.
Jakobson et al. 1995; Sacks 1992; Silverstein 1995), indexes a category that can
loosely be equated with “Americans.” Therefore, it implies that the interviewee
belongs to a group that is assumed to share the stance, which is made relevant
through the action of promising, reported in the question preface, namely as we
have ^promised that, in line 3, we will=, u=h¸<MRC>rid them from this planet</
MRC>, in lines 6–8.11 Moreover, the third-party stance, which by implication
includes the interviewee, is also used as a resource for constructing the actual
question (^How do you ... ^take out a, a ^bin Laden or ‘someone like a bin Laden,
arrows 2). The fact that this stance is here reproduced as an interrogative renders
it presupposed information, i.e., by asking ‘how bin Laden or someone is taken
out’ the interviewer presupposes that ‘somehow it is possible to take bin Laden or
someone out’ (cf. Levinson 1983: 184). This presupposition is also reflected by the
prosodic design of the question TCU: the question word ^How in lines 11 and 12
has primary emphasis, which indeed marks the question to be inquiring about
ways in which bin Laden can be taken out, and not whether it is reasonable or pos-
sible to take him out. By building on the assumption that the interviewee shares
both the stances in the question preface and the presupposition in the question,
the interviewer invites the interviewee to think about measures for ‘taking out bin
Laden,’ and thus raises potential problems for the interviewee.
As we can see in the example, the interviewee does not share the question’s
stances, nor its presuppositions, but designs his answer so that it contains a diver-
gent stance. First, at the very beginning of his answer (arrow 3) the interviewee
displays his divergence from the presupposition by denying it (cf. Haddington
2005b). The divergence is also signaled by the long pause and the dispreference
marker well (cf. Sacks [1973] 1987). After this, the interviewee provides an ac-
count (in line 18) by proposing other alternatives for catching and dealing with
Osama bin Laden. The recycled elements can be seen in the following diagraph.
Diagraph 4 (from example 4): Rid them from this planet

5 IR: we will take them out
12 How you {do} take {a bin Laden} out
17 IE: Well I do n’t know about taking him out Larry
The diagraph shows that in line 17, the recycled forms are again preceded by a
stance marker that contains a first-person subject (I), an auxiliary verb with a
negative marker (don’t), and a (cognitive) verb (know). Moreover, since the index
of the first-person subject I (in line 17) is by implication included in the first-
person-plural we (lines 3, 6, and 10) used earlier, it can be claimed that these
two pronouns resonate with each other, and thus indicate the stance differential
between the interviewer’s and the interviewee’s turns, and the problem that the
interviewee has with the indexical we. As in example (3), the recycled phrase take
them out locates the trouble-source and is used as a resource for producing a
denial to the presupposition in the question. And in line 18, the beginning of the
interviewee’s resolution (But I think we have a lot of options) does not answer the
question in the terms the interviewer set out in the first place, but by new terms
the interviewee himself chooses.
In example (5) below, the interviewee attempts to align with the interviewer’s
positioning question by producing a denial. However, in contrast to example (4),
the interviewer holds the interviewee accountable for not answering the question
by asking the same question again. This example comes from BBC2’s Newsnight.
Here, Jeremy Paxman is interviewing the conservative MP, Tim Yeo, about the
introduction of higher tuition fees in British universities. This issue caused a lot of
dispute and controversy in Britain in spring 2004. The focus is on the question in
lines 24–25, which is a repeated question occasioned by the interviewee’s evasive
answer (data not shown). In other words, the interviewer holds the interviewee
accountable for not answering the question by asking the same question again.
(5) BBC2, Newsnight, Jan 21, 2004: speaking for the universities
IR: Jeremy Paxman, IE: Tim Yeo (037 / 1 / 7:18)
1 IR: Can I ask you just one small factual point,
2 of the eighty-nine Vice Cha=ncellors,
3 (H)(GLOTTAL) Uh,
4 how many of them,
5 .. actually support your position?
24 →1 IR: You keep on claiming to speak for the universities,
25 →1 How many of the Vice Chancellor [s s]upport [2you2].
26 IE: [(GLOTTAL)]
27 →2 [2I2] don’t claim to
28 →2 speak ^for the universities,
29 but [I do claim]--
30 IR: [You’ve been doing no]thing [2^but2] the entire [3ev3]en
31 [4ing4].
32 IE: [2Uh2],
33 [3%3]
34 [4I4] do claim to be concerned about their future,
35 (H) An[d],
The sequential progression of the questioning, which amounts to coercing the in-
terviewee to provide a relevant answer to the question, in itself puts the interviewee
in a difficult position (cf. example 1, above). However, there are also other aspects
that set up a position for the interviewee. First of all, the interviewer’s question,
both the original one in lines 1–5 and the repeated one (arrows 1), presupposes
that at least some Vice Chancellors support the interviewee’s views, which then
projects an answer that should give their exact number. This question is in itself
probably difficult to answer, because not only are the Vice Chancellors numerous,
but it also seems improbable that the interviewee knows exactly which support
him and which do not, if any indeed do. Moreover, the interviewer’s questions are
hostile, because they explicitly hint at a possible discrepancy between the stance
that the interviewee has adopted (in line 24) and the reality (line 25) (cf. Heritage
2003: 81). This is fortified by the design of the question, in which the verb phrase
keep on claiming to (in line 24) suggests that the interviewee’s claims that are now
submitted to critical evaluation have been continual and repeated. All the linguis-
tic and interactional elements in the question intersubjectively set up a position
for the interviewee that he needs to take into account in his response.
In spite of the fact that the interviewer is asking the same question again, the
interviewee denies the assertion (arrows 2) contained in the question in line 24.
As we can see in the diagraph below, the interviewee uses the phrase claiming
to speak for the universities from the interviewer’s question for constructing his
response. Again, the stance marker with a communication verb (I don’t claim)
precedes and is used as a resource together with the recycled unit to engage with
the question, and at the same time to undermine the claim in the interviewer’s
question preface. Note also that the primary stress on ^for in line 28 further em-
phasizes the interviewee’s divergence from the interviewer’s claim.
Diagraph 5 (from example 5): Speaking for the universities

24 IR: You keep on claiming to speak for the universities
27–28 IE: I do n’t claim to speak ^for the universities
However, this time (contrary to example 4, above) the interviewer responds with
a forceful counter-argument. Despite the interviewer’s intervention, the inter-
viewee continues to produce an account of what his actions really are about (line
29). In this case, the interviewee uses the denial + account action combination
as a strategy for evading the hostile question. About four seconds later, the inter-
viewer explicitly voices the fact that interviewee is not answering the question:
(6) BBC2, Newsnight, Jan 21, 2004: speaking for the universities
IR: Jeremy Paxman, IE: Tim Yeo (037 / 1 / 7:51)
46 IR: You’re not going to answer that question [n,
47 IE: [I am going to] answer it,
48 IR: are you].
As we have seen in this and the previous section, there is a strong and recurrent
interactional-linguistic pattern by which interviewees respond to difficult ques-
tions. By using a stance marker combined with a recycled linguistic unit from
the interviewer’s question turn, the interviewee locates a trouble source in the
question (an assertion or a presupposition) and denies it. The trouble source is
located and foregrounded by recycling it in the denial: in examples (1) and (2), a
proposition in the question, in (3), (4), and (5), a presupposition in the question.
The denial, which is composed of the stance marker in association with a recycled
unit from the question, not only sequentially precedes, but also strongly projects
a next action (more evidence for this is given in Section 5). The recycled element
can engage with any element in the question, either in the question preface (Dia-
graphs 1 and 2), the question TCU (Diagraphs 4 and 5) or both (Diagraph 3).
These phenomena are closely related to the interlocutors’ stancetaking and
especially to interviewee alignment with the questions. Indeed, the interviewees
do not just evade the question for the sake of it. Rather, their evasive answers are
contingent products and occasioned by the asserted or presupposed stances in
the question. This practice, therefore, stands as firm evidence of the backward-
type intersubjectivity in news interviews (Haddington 2004). Interviewees recycle
language from the question in order to orient to and undermine a stance encoded
or evoked in the question. The recycled language in this context is used by the
interviewees as a linguistic practice to both orient to and undermine a claim, an
assertion, a presupposition or another problematic position in the question. It is
an extremely efficient and productive practice for the interviewees to avoid the
problematic position established for them in the question but at the same time is
an efficient resource for constituting and organizing a relevant stance in their an-
swer. In the following section, I turn to a more detailed linguistic analysis of what
happens in the second parts of the action combination. I show that interviewees
use a particular turn-constructional format for aligning with preferred, evoked,
and presuppositional stances.
2.4 The neg + pos pattern as a resource for alignment
Interviewees frequently use the stance marker + recycled language, discussed in the
previous section, as a resource for aligning with the interviewer’s difficult question.
This structure recurs in turn-initial position (Haddington 2005b). This is not an
accident, because in that sequential position it has a clear function in the organiza-
tion of the answering turn and thereby also in the organization of the interviewee’s
stance. As was noted, this pattern is used for displaying engagement with the ques-
tion, but also for denying a presupposition or a position set up in the question.
However, denials do not occur alone. Ford (2002) claims that in everyday talk
a denial is actually only one part of a combination of two actions. The second ac-
tion, which follows the denial, is a correction or an account by the same speaker
that gives an alternative interpretation to what has been denied (Ford 2002: 62).
According to Ford (2002), this action combination is a robust and coherent dis-
course structure in everyday talk. This happens in news interviews as well. In my
data, however, this action combination is used exclusively by the interviewees
and, contrary to everyday talk, is frequently produced with the help of a linguistic
pattern called the neg + pos pattern. A search of this pattern in the Santa Barbara
Corpus of Spoken American English (Du Bois et al. 2000; Du Bois et al. 2003; Du
Bois and Englebretson 2004) yielded only one example, which suggests that it is
rare in everyday talk and indicates that it has a special function in news inter-
views. By using it, the interviewees can design their turn so that rather than just
bluntly evading the presuppositions or the preferred stances encoded and evoked
in the questions, they actually align with them. The following diagraphs depict the
moments in which the interviewees first produce the denying actions and then
proceed to give an account or a resolution for the denial. The bolded parts picture
the linguistic pattern.12
Diagraph 6 (from example 1): Bomb him out

7 IE: I d-
8 I do n’t know that you could bomb him out
9 I mean
10 I I think that
12 the magnitude of this operation
Diagraph 7 (from example 2): Rid them from this planet

17 IE: Well I don’t know about taking him out Larry
18 But I think we have a lot of options
Diagraph 8 (from example 3): The new bin Laden tape

28 IE: I don’t think this is a matter of cryptography
29 I think this is a question of whether
Diagraph 9 (from example 4): Cheap air travel

22 IE: No I do- I-
23 Now I would --
24 I wouldn’t argue that it’s too cheap,
25 I do think though,
26 it ought to meet the costs {}
Diagraph 10 (from example 5): Speaking for the universities

4 IE: I do n’t claim to speak for the universities
6 but I do claim
9 I do claim to be concerned about their future
The neg + pos pattern not only plays an important role in the organization of
interviewees’ multi-unit responses, but also organizes the interviewee’s intersub-
jective stancetaking. It is composed of two parts, which always occur in different
intonation units. Quite often, but not always, these two parts also follow each
other in consecutive intonation units.13 The first part doing the denial contains
a first-person-singular pronoun (or sometimes some other pronoun) and a ne-
gated cognitive or communication verb, which is followed by a recycled linguistic
element that identifies the issue that is being denied. The second part doing the
account, then contains a first-person-singular pronoun with a cognitive or com-
munication verb, which is followed by the account. The pattern can be schema-
tized in diagraph form as:
“I” + negative particle + cognitive / communication verb + recycled language
“I” + verb + predication
It is worth noting that the original linguistic unit that the interviewee recycles is
rarely negatively formulated. This suggests that this practice in this interactional
context can only be used if the unit that is recycled does not contain negative
markers. One reason for this might be that the denial that projects the account al-
ways contains a negative marker and that a positively formulated utterance in this
context is easier to recycle. However, in cases in which the recycled linguistic unit
comes from a negatively formulated TCU, the recycled element can be slightly
modified. Consider the following deviant case:
(7) BBC2, Newsnight, Jan 19, 2004: Keep a promise

IR: Jeremy Paxman, IE: Tony Blair (036 / 1 / 3:34)
1 IR: At `what ^point,
2 after ^writing that `manifesto,
3 in which <MRC>you requested our votes</MRC>,
4 did you `realise,
5 →1 (H) that you couldn’t keep the promise?
6 IE: (TSK)(H) ... Well I,
7 →2 <A>I ^don’t àccept that we have broken the promise,
8 (0) as I said</A>,
9 because the `new system doesn’t come into êffect,
10 .. until âfter the next general `%election.
In the above example, the interviewee recycles the structure of the phrase keep
the promise (arrow 1) in the interviewer’s question but changes it into broken the
promise (arrow 2), which, of course, communicates an opposite meaning com-
pared to the original phrase. Consider the following diagraph:
Diagraph 11 (from example 7)

5 IR: that you could n’t keep the promise
7 IE: I don’t accept that we have broken the promise
This change shows that the denial + account action combination has a strong
internal organization, and that the stance marker in the denial is a “compulsory”
part of the combination and can, in fact, influence the design of the recycled lin-
guistic elements in the complement.14
The neg + pos pattern does not, of course, only occur in this interactional
context, but it is a recurrent and routinized pattern in political news interviews
and reproducible in this interactional context. By using this particular pattern,
the interviewees deny an assertion or a presupposition in the interviewer’s ques-
tion and then provide the next relevant action, an account for the denial. It is
worth noting that even though this pattern occurs in environments in which the
interviewee does not produce a preferred answer to the question, the interviewees
do not just bluntly avoid answering the question, but in fact engage with it. Con-
sequently, this pattern shows that the interviewee orients to the position (or some
part of it) set up in the question, undermines it and goes on to propose another
way of looking at the issue, i.e., she aligns with the question.
Following Ford (2002), this action combination and the linguistic pattern in
it could be perceived as a turn-constructional format, a pattern in which two ac-
tions are closely connected. As Ford (2002) claims, a denial alone strongly proj-
ects a resolution or an explanation component. This is particularly true in news
interviews in which denial alone would be perceived as inadequate, because inter-
viewees are expected to support and give grounds to their statements. Therefore,
the denial projects a move toward the second part, a resolution or a counterstance.
Consequently, these two actions are not just individual actions, but they organize
a larger intersubjective stancetaking activity in which the interviewee responds to
an evoked, asserted, or presupposed stance in the question.
Further evidence for the connectedness of the two actions can be heard in
how this pattern is designed prosodically. In all examples except (1) (see Figures
2–5), the second part of the action combination is latched onto the first part. In
other words, in order to keep the turn, the interviewee produces a rush-through
(Local 1992; Schegloff 1982), which helps him get to the next TCU. In example
(1), the interviewee produces a clearly audible inbreath, which also projects more
talk to come (see Figure 1). In addition to this, as the acoustic measurements
suggest, in some cases (see the pitch curve in Figures 2 and 5), the end of the first
part is produced with distinctly rising-continuing intonation, which also projects
further talk from the speaker. These phonetic features further suggest that the
first part projects a move toward the second part. The interviewer can use these
features as a resource for understanding what the interviewee is doing in terms
of turn projection, i.e., that the interviewee is going to produce more talk beyond
the next TRP.
Although these features display the close relationship between the two TCUs
within the interviewee’s turn, they also serve an interactional function. As was
noted above, interviewers have the right to interrupt or challenge the interviewee
if they feel that she is not answering the question. Thus, it is possible that when
an interviewee produces a denial, the interviewer can intervene after the denial
and challenge or disagree with the interviewee’s denial. This actually happens
in example (5) in line 30, when the interviewer says [You’ve been doing no]thing
[2^but2] the entire [3ev3]en[4ing4]. By producing the action combination pro-
sodically as they do (no pause, rising intonation, and latching onto previous unit),
the interviewees minimize the interactional space at the TRP in which the inter-
viewer has an opportunity to intervene. Moreover, as we can see in example (5),

the interviewee continues and finishes the second part of the action combination,
which shows that despite the interviewer’s intervention, this turn-internal orga-
nization of the two actions is very strong.
Another interesting point that arises in these examples is that there is a ten-
dency in how this pattern is produced prosodically. In all of the examples above,
the maximum pitch in the first part is produced on the negative modal or aux-
iliary verb (either don’t or wouldn’t). In addition to this, the unit that is recycled
from the interviewer’s question also receives strong emphasis. Furthermore, in
the second part, the primary emphasis tends to be on the first (cognitive or com-
munication) verb or an accompanying modal. These similarities in the prosodic
design of this action combination act as evidence of intersubjective stancetaking,
i.e., the interviewee is taking a stance based on an assertion or presupposition in
the question turn, and thereby aligning with it. In other words, the practices in
the action combination and the action combination itself are contingent achieve-
ments made relevant because of the position set up in the question.
As was claimed above, this linguistic pattern almost never occurs in everyday
talk. Therefore, an interesting question arises: What motivates the use of this pat-
tern in news interviews? Sacks ([1973] 1987) noted that questions tend to prefer
agreement. He also noted that if answerers treat questions as problematic and show
this by using various dispreference markers, questioners tend to modify the ques-
tion so that it contains the opposite preference. In other words, the questioners
end up using a form that answerers can more readily agree with. However, in news
interviews the interviewers rarely modify their questions for the interviewees.
Rather, the whole issue of preference is turned upside-down. As we have seen, the
Figure 1. Prosody in example (1): Cheap air travel

Figure 2. Prosody in example (2): Bomb him out
Figure 3. Prosody in example (3): The new bin Laden tape
Figure 4. Prosody in example (4): Rid them from this planet
Figure 5. Prosody in example (5): Speaking for the universities
interviewers can deliberately position their guests by designing their questions so

that they prefer a problematic stance. Since the interviewees are forced to answer
the questions, they have to align with the problematic stance in that question im-
mediately. Because of the pressure, the above linguistic pattern is a useful resource
for the interviewees to align with the question. In other words, they can produce
a stance that diverges from but still does not totally disagree with an assertion or
a presupposition in the question. By designing the answer in this way, they can
reduce the severe consequences that a strict disagreement might bring with it.
3. Discussion and conclusions
News interviews are publicly broadcast interactions in which the journalist’s pri-
mary task is to ask questions of a public figure. The public figure is then expect-
ed to provide appropriate answers to the questions (e.g., Scannell 1991: 4). One
purpose of this question-answer activity is to give information, express opinions,
and discuss and debate topical and often controversial issues. Against this back-
ground, news interviews are the venues for politicians and other experts to pub-
licly convey, formulate, defend, and negotiate their stances.
This paper makes the following claims about stancetaking in news interviews.
First of all, it shows that the methodological approach to study stancetaking with
data from real interactional situations is able to provide new and interesting find-
ings about the ways in which speakers take stances. It also shows that speakers
do not (just) express their subjective stances, but indeed orient to and engage
with each other’s stances, and thus engage in intersubjective stancetaking. It has
also shown that a linguistic analysis combined with the interactional analysis of
the data can provide new findings about stancetaking. The linguistic analysis es-
pecially shows that participants in news interviews rely on recurrent linguistic
formats and patterns in specific interactional situations, for example in producing
the denial + account action combination.
Second, this paper has described some practices by which interviewers set up
positions for the interviewees and how the interviewees align with these positions.
The positioning activity is essentially a one-directional and forward-type inter-
subjective activity. By setting up a position the interviewers attempt to constrain
and delimit the possibilities for the interviewee to construct a responsive stance.
It is worth noting, however, that not all questions position the interviewees. Of-
ten (but not always) simple information-seeking wh-questions position the inter-
viewees only minimally, if at all, because they establish a fairly loose agenda and
project a broader answer than, for example, negative interrogatives or tag ques-
tions, which have a fairly clear preference structure (cf. Heritage 2002, 2003: 69).
It is also important to notice that positioning becomes clearly visible only if the
interviewee treats the questions as doing positioning. In some cases, the inter-
viewer may be seen to be positioning the interviewee, but the interviewee does
not treat the question as problematic. The interviewees indeed can decode pre-
ferred stances and presuppositions in the questions and orient to and deny them.
One practice for denying an element in the question is the above-mentioned de-
nial + account action combination in which interviewees use the stance marker
(usually personal pronoun + negative particle + cognitive/communication verb)
together with recycled language. As was shown, this linguistic pattern is a routin-
ized and reusable linguistic pattern in this interactional context.
Third, these activities are made possible by three special features of news in-
terview interaction. First, the special (institutional) turn-taking organization in
which sequences of questions and answers follow each other. Second, the turn-
type pre-allocation, i.e., the fact that interviewers produce questions and inter-
viewees produce answers means that only interviewers do the positioning and
the interviewees the aligning. Third, the long multi-unit turns – which provide
for the possibility that individual TCUs within a turn begin to resonate with each
other – produce action combinations and combinations of linguistic practices
that become relevant for the production of these activities.
There are also some general observations that can be drawn from the analysis
above. First of all, recent research on stancetaking, attitudes and evaluation sug-
gests that participants in everyday conversation do not converse with particular
pre-determined stances in their heads, but rather work together in interaction in
order to reach some sort of a joint and negotiated stance (Du Bois this volume;
Kärkkäinen 2003b; Potter 1998). However, the findings in this paper suggest that
stancetaking is different in news interviews. As was shown above, interviewers
use various practices and actions – ranging from the use of particular pronouns
to different types of action combinations – for putting the interviewees between
a rock and a hard place. The way in which these practices, actions, and the com-
binations of them function in the question turns suggests that interviewers come
to the interview situation with particular questions and agendas in mind. They
use their knowledge of who the interviewees are, what their background is, who
or what they represent, and what kind of answers can be expected from them, for
constructing their questions so that the answers the interviewees would prefer are
very hard to implement.
The idea of positioning as an intersubjective activity also forces us to recon-
sider the idea of interviewee evasion. Not only are evasive or agenda-shifting ac-
tions often occasioned by the interviewers’ questions and therefore intersubjec-
tive, but they are also contingent achievements occasioned by the evoked and
preferred stances and presuppositions in the questions. Consequently, it is better
not just to assume that interviewees evade questions, but rather to talk about an
aligning activity by which interviewees attempt to find a place to carefully word
their stances that not only take into account the evoked and presupposed stanc-
es, but also reflect the speaker’s own identity, background, aims, and previously
stated stances. The linguistic pattern discussed above stands as good evidence of
this. The interviewees use this pattern for vitiating preferred stances and presup-
positions, and for expressing the stances that better represent their views of the
topical matter at hand, i.e., views that they have very likely had before they came
to the program. We could therefore call into question whether the stances in news
interviews are as emergent as they are in everyday talk, although the design of the
interviewee’s stance is obviously contingent upon the design of the question. On
the contrary, it seems that the interviewees do not assume and accept the posi-
tions and stances put on the table by their host but firmly state and support their
own stances.
It seems that stancetaking in news interviews relies on some types of prior
beliefs, or contextual, political, and cultural values that are bound to the chosen
topical agendas and the viewpoints assumed by the participants. In other words,
stancetaking in news interviews is perhaps even more about ‘who you are,’ ‘what
you say,’ and ‘how you say it,’ than it is in everyday talk.
The stancetaking activities described above also have interesting implications
for the audience. As Heritage (1985) has shown, for example, the fact that in-
terviewers sometimes reformulate interviewee answers in news interviews shows
that news interviews are produced for an overhearing audience. It would seem
that the stancetaking activities of positioning and alignment are also targeted for
the audience in that they produce and distribute information, create entertain-
ment, report opinions and political viewpoints, tell news for the audience, and
thereby also (attempt to) influence and shape public opinions. In fact, these ac-
tivities are characteristic of news interviews, but are not likely to occur in other
forms of question-driven institutional interaction (e.g., doctor-patient interac-
tion). It seems, therefore, that it is not the institutionality per se that occasions
these activities, but the fact that news interviews are publicly broadcast programs
and aired for an audience. Although it is quite possible that one rationale behind
producing these activities is the entertainment value of news interviews, we can
only speculate about the significance of news interviews as popular media and
their function in and impact on society and culture as a producer and bearer of
particular beliefs and value-systems. However, it would seem likely that on some
level these situated activities may have an impact on how the individuals in the
audience form their opinions about the topics in news interviews.
Notes
1. I am greatly indebted to the following persons for help, comments and encouragement:
John Du Bois, Robert Englebretson, Tiina Keisanen, Elise Kärkkäinen, Maarit Niemelä, Arja
Piirainen-Marsh, and Mirka Rauniomaa. I also want to thank the anonymous reviewer for
helpful comments. Naturally, I alone am responsible for any errors and inadequacies that re-
main. This research has been partly funded by a grant from the Academy of Finland (research
project 53671).
2. Turn constructional unit (TCU) is the basic unit out of which speakers set out to construct
talk (Sacks et al. 1974). One TCU can constitute a recognizably complete turn and can char-
acteristically be lexical, phrasal, sentential, or clausal. These linguistic characteristics of turn
construction provide for “projectability of a turn,” i.e., the possibility for co-participants to real-
ize and understand what is under way and to project the possible completion point of a TCU
(Sacks et al. 1974). In everyday conversation, each speaker gets the right to construct a single
TCU to a possible completion. In institutionalized interaction, however, the allocation of turns
is pre-determined.
3. TRPs refer to the ends of TCUs, i.e., that there is a possibility for transition between speakers
at the end of a TCU (Sacks et al. 1974). However, in news interviews, TRPs are more “neutral”
than in everyday talk (cf. Schegloff 1996: fn. 14; 2001). This is a consequence of the ‘multi-unit-
ness’ of turns and the pre-allocated turn types in news interviews. For example, interviewees
are entitled to and even assumed to give answers that contain more than one TCU. Therefore,
the turn-taking rules that apply to everyday talk (Sacks et al. 1974), in which one speaker is
basically allotted one TCU, and in which different rules of “current speaker selects next” or
“self-selection” apply at each TRP, do not apply to news interview interaction in the same way
(see for example Clayman and Heritage 2002; Greatbatch 1988; Heritage 1985).
4. Although there are only scant mentions of how the interviewer’s question acts as an impe-
tus for the interviewee’s evasiveness, Nuolijärvi (1994) discusses how the interviewer’s turn in
fact creates the foundation and thereby acts as the sequential impetus for the interviewee’s non-
answers, and Clayman and Heritage (2002: 188–237) address the issue occasionally, but only in
passing, together with the analysis of examples.
5. Note that the talk and the question in lines 1–3 is not part of the actual interview, but added
in the program after the actual interview. Nevertheless, I treat the reported speech as faithful to
the actual question in the interview.
6. Editing is used frequently in documentaries and news reports.
7. Diagraph is a technical term used in dialogic syntax (Du Bois 2001, this volume). Dia-
graphs are used to depict and illustrate linguistic relations between the talk of two speakers or
within a speaker’s turn. One line in a diagraph represents one intonation unit. The purpose of
diagraphs here is to show how speakers use each other’s syntax and other linguistic material to
construct their stances. Curly brackets in the diagraphs indicate that some unit has been moved
for representational purposes.
8. See more about the notion of neutralism and how it relates to stancetaking in Haddington
(2004).
9. I thank Marja-Leena Sorjonen for this observation.
10. Example (3) is analyzed in detail in Haddington (2004).
11. In addition to seeing this as a promise that is made to the allies of United States, it is also
possible to see this action as involving an (indirect) threat to al-Qaida.
12. A more quantitative analysis of this pattern in this action combination is given in Had-
dington (2005b). It also discusses the different types of denials and the ‘claim for insufficient
knowledge + explanation’ action combination that occur in news interviews.
13. It seems that this pattern is used in this order particularly for denying something in the
interviewer’s question. Other orders also occur and their analysis may reveal other functions.
This, however, requires further research.
14. For more detailed discussion, see Haddington (2005b).
References
Berg, M. 2001. “Vallan vahtikoira haukkuu mutta ei pure: Poliitikkoa syyllistävät toimittajan
kysymykset televisiokeskustelussa [The watchdog of power barks but does not bite: Inter-
viewer’s accusatory questions toward politicians in television conversation].” In Institu-
tionaalinen vuorovaikutus: Keskustelunanalyyttisia tutkimuksia [Institutional interaction:
Conversation analytic studies], J. Ruusuvuori, M. Haakana and L. Raevaara (eds.), 121–137.
Helsinki: Suomalaisen Kirjallisuuden Seura.
Berg, M. 2003. Syytöksiä ja epäilyjä: Toimittajan ja poliitikon vuorovaikutuksesta televisiokes-
kustelussa [Accusations and doubts: Interaction between journalists and politicians in televi-
sion conversation]. Helsinki: Suomen Kirjallisuuden Seura.
Biber, D., Conrad, S. and Leech, G. 2003. Longman Student Grammar of Spoken and Written
English. Essex: Longman.
Chafe, W. 1986. “Evidentiality in English Conversation and Academic Writing.” In Evidential-
ity: The Linguistic Coding of Epistemology, W. Chafe and J. Nichols (eds.), 261–72. Nor-
wood, NJ: Ablex Publishing Corporation.
Clayman, S.E. 1988. “Displaying neutrality in television news interviews.” Social Problems 35
(4): 474–492.
Clayman, S.E. 1993. “Reformulating the question: A device for answering/not answering ques-
tions in news interviews and press conferences.” Text 13: 159–188.
Clayman, S.E. 2001. “Answers and evasions.” Language in Society 30: 403–442.
Clayman, S.E. and Heritage, J. 2002. The News Interview: Journalists and Public Figures on the
Air. Cambridge: Cambridge University Press.
Pennsylvania.
Du Bois, J.W., Chafe, W., Meyer, C., Thompson, S.A., and Martey, N. 2003. Santa Barbara Cor-
pus of Spoken American English, CD, Part 1. Philadelphia: Linguistic Data Consortium,
University of Pennsylvania.
DVD, Part 3. Philadelphia: Linguistic Data Consortium, University of Pennsylvania.
Du Bois, J.W., Schuetze-Coburn, S., Cumming, S. and Paolino, D. 1993. “Outline of Discourse
Transcription.” In Talking Data: Transcription and Coding in Discourse Research, J.A. Ed-
wards and M.A. Lampert (eds.), 45–89. Hillsdale, NJ: Lawrence Erlbaum Associates.
Ford, C.E. 2002. “Denial and the construction of conversational turns.” In Complex Sentences
in Grammar and Discourse. Essays in honor of Sandra A. Thompson, J.L. Bybee (ed.) 61–78.
Philadelphia: John Benjamins.
Ford, C.E. 2004. “Contingency and units in interaction.” Discourse Studies 6 (1): 27–52.
Goffman, E. 1981. “Footing.” In Forms of Talk, E. Goffman (ed.) 124–159. Philadelphia: Univer-
sity of Pennsylvania Press.
Greatbatch, D. 1986. “Aspects of topical organization in news interviews: the use of agenda–
shifting procedures by interviewees.” Media, Culture & Society 8: 441–455.
Haddington, P. 2004. “Stance taking in news interviews.” SKY Journal of Linguistics 17: 101–
142.
Haddington, P. 2005a. The Intersubjectivity of Stance Taking in Talk-in-interaction. Ph.D. dis-
sertation. Oulu: Oulu University Press.
Haddington, P. 2005b. “The linguistic NEG + POS pattern and two action combinations as
resources for interviewee stance taking in news interviews.” In Kieli ja sosiaalinen toim-
inta [Language and Social Action], L. Kuure, E. Kärkkäinen and M. Saarenkunnas (eds.),
85–107. AFinLA Yearbook. Publications de l’association finlandaise de linguistique appli-
quée.
Heritage, J. 1984. Garfinkel and Ethnomethodology. Cambridge: Polity Press.
Heritage, J. 1985. “Analyzing news interviews: Aspects of the production of talk for an ‘over-
hearing’ audience.” In Handbook of Discourse Analysis, Vol. III, T.A.V. Dijk (ed.) 95–119.
London: Academic Press.
Heritage, J. 2002. “The limits of questioning: Negative interrogatives and hostile question con-
tent.” Journal of Pragmatics 34 (10–11): 1427–1446.
Heritage, J. 2003. “Designing questions and setting agendas in the news interview.” In Studies in
Language and Social Interaction: In Honor of Robert Hopper, P. Glenn, C.D. LeBaron and J.
Mandelbaum (eds.), 57–90. London: Lawrence Erlbaum.
Heritage, J. and Greatbatch, D. 1991. “On the institutional character of institutional talk: The
case of news interviews.” In Talk and Social Structure, D. Boden and D. Zimmerman (eds.),
93–137. Oxford: Polity Press.
Heritage, J.C. and Roth, A.L. 1995. “Grammar and institution: Questions and questioning in
the broadcast news interview.” Research on Language and Social Interaction 28 (1): 1–60.
Horn, L.R. 1989. A Natural History of Negation. Chicago: The University of Chicago Press.
Hutchby, I. and Wooffitt, R. 1998. Conversation Analysis. Cambridge: Polity Press.

Jakobson, R., Waugh, L.R. and Monville-Burston, M. 1995. On Language. Harvard: Harvard
University Press.
Karlsson, S. 2003. “Interactional uses for an epistemic marker: The case of ‘jag tycker’/‘tycker
jag’ in Swedish.” Melbourne Papers in Linguistics and Applied Linguistics 1: 5–23.
Kärkkäinen, E. 2003a. Epistemic Stance in English Conversation: A Description of Its Interac-
tional Functions, with a Focus on ‘I think’. Amsterdam: John Benjamins.
Kärkkäinen, E. 2003b. “‘Is she vicious or dense?’: Dialogic practices of stance taking in conver-
sation.” In Functional Approaches to Grammar: Theoretical and Methodological Consider-
ations., H. Tao (ed.) 48–66. Santa Barbara Working Papers: University of California, Santa
Barbara.
Levinson, S.C. 1983. Pragmatics. London: Cambridge University Press.
Local, J. 1992. “Continuing and restarting.” In The Contextualization of Language, P. Auer and
A. di Luzio (eds.), 273–296. Amsterdam: John Benjamins.
Nuolijärvi, P. 1994. “Talousrikollinen televisiohaastattelussa: Vastauksen muotoilemisen prob-
lematiikkaa [White-collar criminal in a television interview: Problems in designing an an-
swer].” In Talous ja kieli II [Economics and Language II], M. Majapuro and T. Nikko (eds.),
228–244. Helsinki: Helsinki School of Economics and Business Administration.
Potter, J. 1998. “Discursive social psychology: From attitudes to evaluative processes.” European
Review of Social Psychology 9: 233–266.
Sacks, H. [1973] 1987. “On the preferences for agreement and contiguity in sequences in con-
versation.” In Talk and Social Organization, G. Button and J.R.E. Lee (eds.), 54–69. Clev-
edon: Multilingual Matters Ltd.
Sacks, H. 1992. Lectures on Conversation [1964–1968]. Cambridge: Blackwell Publishers.
Sacks, H. and Schegloff, E.A. 1979. “Two preferences in the organization of reference to per-
sons and their interaction.” In Everyday Language: Studies in Ethnomethodology, G. Psathas
(ed.). New York: Irvington Publishers.
turn-taking for conversation.” Language 50: 696–735.
Scannell, P. 1991. “Introduction: The relevance of talk.” In Broadcast Talk, P. Scannell (ed.),
1–13. London: Sage.
Schegloff, E.A. 1982. “Discourse as an interactional achievement: Some uses of ‘uh huh’ and
other things that come between sentences.” In Georgetown University Roundtable on Lan-
guages and Linguistics 1981; Analyzing Discourse: Text and Talk, D. Tannen (ed.), 71–93.
Washington D.C.: Georgetown University Press.
Schegloff, E.A. 1996. “Turn organization: One intersection of grammar and interaction.” In In-
teraction and Grammar, E. Ochs, E.A. Schegloff and S.A. Thompson (eds.), 52–133. Cam-
Scheibman, J. 2001. “Local patterns of subjectivity in person and verb type in American Eng-
lish conversation.” In Frequency Effects and Emergent Grammar, J. Bybee and P.J. Hopper
Silverstein, M. 1995. “Shifters, linguistic categories and cultural description.” In Language, Cul-
ture and Society, B.G. Blount (ed.), 187–222. Prospect Heights, IL: Waveland Press, Inc.
Simon-Vandenbergen, A.-M. 2000. “The functions of I think in political discourse.” Interna-
tional Journal of Applied Linguistics 10 (1): 41–62.
ten Have, P. 2001. “Applied conversation analysis.” In How to Analyse Talk in Institutional Set-
tings: A Casebook of Methods, A. McHoul and M. Rapley (eds.), 3–11. London: Continu-
um.
Thompson, S.A. 2002. “‘Object complements’ and conversation: Towards a realistic account.”
Studies in Language 26 (1): 125–164.
Based on Du Bois et al. (1993).
Speech
Units
Intonation unit {line break}
Truncated intonation unit ––
Truncated word – (en dash)
Final .
Continuing ,
Appeal (seeking a validating response from listener) ?
Speakers
Speech overlap [ ]
(numbers inside brackets index overlaps) [2 two words 2]
Primary accent (prominent pitch movement carrying
intonational meaning) ^
Secondary accent `
Unaccented
Lengthening =
Pause
Long pause (0.7 seconds or longer) ...(N)
Medium pause (0.3 – 0.6 s) ...
Short (brief break in speech rhythm)(0.2 or less) ..
Latching (0)
Vocal noises
Alveolar click (TSK)
Glottal stop (GLOTTAL)
Inhalation (H)
Laughter during speech (1–5 words) @
(e.g., @two @words)
Quality
Tempo and rhythm
Allegro: rapid speech <A> </A>
Lento: slow speech <L> </L>
Marcato: each word distinct and emphasized <MRC> </MRC>
Rhythmic: stresses in a beatable rhythm <RH> </RH>
Voice quality
Creak %
Creak during speech %
(e.g., %two %words)
Uncertain hearing #
(e.g., #two #words)
Indecipherable syllable #
Specialized notations
Restart {Capital initial}
Name index
A F M
Antaki, Charles 52 Field, Margaret 17 MacWhinney, Brian 120
Athanasiadou, Angeliki 16 Finegan, Edward 17 Martin, J. R. 16
Fitzmaurice, Susan 19 Matoesian, Gregory 18
B Ford, Cecelia E. 284, 303, 306 McConnell-Ginet, Sally 134
Bailey, Guy 54 Milroy, James 53
Becher, Tony 35–36 G Milroy, Lesley 53
Bednarek, Monika A. 16 Gilman, Albert 74 Mulac, Anthony 190
Benveniste, Émil 16, 122, 125 Groom, Nick 35–36
Benwell, Bethan 83 N
Berman, Ruth 18, 111–113, 118 H Nuyts, Jan 125
Biber, Douglas 16–17, 28–30 Haddington, Pentti 22, 194
Bouacha, M. A. 118 Hall, Kira 52, 84 O
Brown, Roger 74 Hazen, Kirk 54 Ochs, Elinor 18, 51
Bucholtz, Mary 52, 84 Helasvuo, Marja-Liisa 225–226
Heritage, John 255–256, 284, P
C 297, 311 Pomerantz, Annita 277
Chafe, Wallace L. 120, 190 Hunston, Susan 16, 20, 39–44, Precht, Kristen 18
Channell, Joanna 36–37 123
Charles, M. 28, 30–34 Hyland, Ken 17, 20 R
Clayman, Steven 284, 297 Rampton, Ben 54
Clift, Rebecca 184, 260 J Rauniomaa, Mirka 21
Conrad, Susan 28–30 Johnstone, Barbara 18, 20, 55 Rose, Mary 54
D K S
Dasher, Richard 125 Kafura, Dennis 116–117, 134 Sacks, Harvey 256, 290–291,
Du Bois, John W. 20, 51–52, Kärkkäinen, Elise 17–18, 21, 307
140, 199, 225, 231, 283, 285– 183, 186, 189, 226, 292 Schegloff, Emanuel A. 256,
286 Keisanen, Tiina 21 284, 289–291
Duranti, Alessandro 97, Kiesling, Scott F. 55 Scheibman, Joanne 17–18, 21,
103–104 Koshik, Irene 254–256 187–188
Schiffrin, Deborah 125, 188–
E L 189, 207–209
Echols, John M. 76 Labov, William 53 Schilling-Estes, Natalie 54
Eckert, Penelope 55 Langacker, Ronald W. 16, 115, Shadily, Hassan 76
Edwards, Derek 19 120 Shoaps, Robin 18
Englebretson Robert 18, 20–21, Lenk, Uta 189 Simon-Vandenbergen, Anne-
72–73 LePage, Robert B. 53–54 Marie 289
Errington, J. Joseph 74 Louw, Bill 37–39, 41–42 Sinclair, John 37, 39
Lyons, John 15–16, 118 Sneddon, James N. 76–77
Stokoe, Elizabeth 83
320 Name index
Strauss, Claudia 125 Trudgill, Peter 53 Wittgenstein, Ludwig 171

Stubbs, Michael 17, 70 Tse, Polly 17, 20 Wouk, Fay 70
Wu, Ruey-Jiuan Regina 18,
T V 88–89, 95, 225
Tabouret-Keller, Andrée 53–54 van Leeuwen, Theo 111, 117–118
Thompson, Geoff 16, 123 Verhagen, Arie 19, 122 Z
Thompson, Sandra A. 187, Zhang, Qiao 111
189, 213 W
Traugott, Elizabeth Closs 125, White, Peter R. R. 16, 19
188 Widdicombe, Susan M. 52
Subject index
A complement-taking predicates F
adjacency pairs 256, 290–291 17, 30, 186–189 finite clauses 29
adjectives 11–14, 33–36 constructed dialogue 85–86 first–person reference (in
adverbs 17, 29, 91, 188 Conversation Analysis 19–20, Colloquial Indonesian) 73–
stance adverbials 29–30, 71 184, 253–254, 283–285 88, 104–105
affect see affective stance corpus linguistics 20, 27–28
affective stance 17, 94–95, 105, corpus of conversational Finnish G
143–144, 152, 162, 260–261, (Helsinki) 223–224 gaze 195–196
272 generality of reference 111–117
affiliation 54, 70, 74–75, 197 D general subjects 118–122, 126
see also solidarity diagraph 160–161, 166, 172, generic you 130–133, 296
agency 97–99, 103–105 199–200, 231 grammar pattern 30–34
alignment 139, 144–145, 150– dialect identity see local
153, 159, 162–167, 169–170, identity H
197, 226, 253, 277, 283, 285, dialogicality 19, 139–141, 147, hedges 222
310 152, 157–158, 160–161
the neg + pos pattern 303– see also intersubjectivity I
309, 311 dialogic syntax 140, 160, 172 identity 50–55, 73–74, 83,
to hostile interview questions diathesis (in Colloquial 87–88
287–290, 296–297 Indonesian) 95–105 performance of 51–55,
to preferred stances 293– discourse markers 188–189, 62–66, 83–84
297 206, 208–211 see also local identity
to presuppositions 299–303 discourse stance 18–19, 111–113 I guess (English)
intersubjectivity in second–position actions
appraisal 142 E 197–203
assessment of interactional epistemicity see epistemic in sequence–initiating
relevance 93–95, 105 stance actions 190–197
assessments 142, 225–226 epistemic phrases 186–190, 212 in side sequences 204–208,
attitudinal stance 29, 36 epistemic stance 17–18, 20–21, 210
29, 50–51, 61–66, 71, 88–89, inference 191–194
B 91–95, 105, 143–144, 156, 162, institutional stance see moral
British National Corpus 4–6, 183, 222–223, 253, 257 stance
11–13, 22–23 evaluation 6, 9–10, 12–14, interactional linguistics 253–
and VIEW (Variation in 16–17, 35–36, 40–41, 112–113, 254
English Words and 122–124, 127–129, 141–146, interrogatives
Phrases) 4, 22–23 153, 155, 158, 163–166, 169–170, negative yes/no questions
191–194, 226, 253, 270 255–257
C evaluative stance see as challenge to prior
challenges 254–255 evaluation action 271–273
collocations 11–15, 188–189 evidentiality 91–93, 95, 105 as challenge to telling
see also phraseologies evidentials 190, 211–212 sequence 266–269
complement clauses 17, 30, 188
322 Subject index
tag questions 255–257, 310 phraseologies 27, 36–37, 39–40, consequentiality of 6–8, 11,
as challenge to prior 42–45 14–15, 173
action 269–271, see also collocations definitions and
273–275 physical stance 6–8, 12, 14, 18 characterizations of 1–4,
as challenge to telling positioning 95–99, 103–105, 6–15, 21, 51–52, 142–145
sequence 258–266 113–114, 139, 143, 145, diversity of 1–4, 15, 144–145,
wh-questions 254–255, 310 152–153, 155, 158–159, 169
intersubjectivity 19–20, 111, 162–166, 169–170, 283, 286, historical development of
113–114, 124–125, 133–134, 310 18–19
140–141, 158–162, 168, 170, by hostile questions 287– indexicality of 6, 8, 11,
184, 212, 283, 286, 309–311 290, 295–296 14–15, 18–19, 51, 55, 111–112,
see also alignment by preferred stances in 114, 124, 146
questions 290–292, interactional nature of 6,
L 294–297 8, 10–11, 13–15, 19–22,
local identity 20, 49–50, 53–56, by presupposition in 112–114, 125–126, 128–129,
62–66 questions 297–301 134, 157, 171, 185, 254,
preference structure 290–291 277–278, 283
M prepositional phrases 29 interdisciplinary nature of
meaning group 30–34 presupposition 298–300 1–4
minun mielestä and minusta prosody 10, 260–261, 263, in writing 19–20
(Finnish) 270–272, 307–309 joint construction of see
marking transition to a first stance: interactional
assessment across turns Q nature of
233–237 questions see interrogatives metalinguistics of 3–15
marking transition to first projectability of 151
assessments within R public nature of 6–8, 10,
extended turns 237–241 register 6, 14, 29–30 13–15, 163, 169, 171, 173
opening up assessments for reported speech see relational nature of see
mutual attention 241– constructed dialogue stance: interactional
246 nature of
position in IU 226–227 S responsibility for 147, 171,
position in turn sequence Santa Barbara Corpus of Spoken 173, 277–278
227–228 American English 4–10, 22, sequential nature of 283
projecting disagreement 255–257 sociocultural dimensions
in second assessments semantic prosody 39 of 6, 8, 11, 15, 18, 43–44,
228–233 sequence organization 277–278 112–114, 117–118, 124–126,
modals 17, 71 sequentiality 149, 154–156, 128–129, 134, 139–141, 173
moral stance 6, 8, 10, 11–14, 18, 159–161, 168, 212 see also affective stance,
96–99, 103–104 social network theory 53 appraisal, attitudinal
sociolinguistics (variationist) stance, discourse stance,
N 52–54, 83 epistemic stance,
news interviews 255–256, 284, solidarity 114–115, 124, 126, evaluation, moral stance,
309–312 128–129 physical stance, style
non-personal they 127–129 see also affiliation stance
-nya clitic (Colloquial stance stance differential 166–167, 170,
Indonesian) 88–95, 105 across genre 19 225, 300
and first-person subjects stance frame 184–185, 188
P 225 stance object 143, 147–148,
particles 70 as collaborative activity 151–157, 159, 163–166, 169–170,
personal stance see evaluation see stance: interactional 283
nature of
Subject index 323
stance subject 147, 151–153, 155, T V

157, 159, 163–166, 169–170 transcription 4, 20 voice (grammatical) see
style stance 29 turn design 194–197, 277–278 diathesis
subjectivity 15–18, 111, 113, 122,
125, 133–134, 140–141, 152–154, U
156–157, 159, 170, 184, 225 usage-based linguistics 4,
69–71
Pragmatics & Beyond New Series
A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com
171 Félix-Brasdefer, J. César: Politeness in Mexico and the United States. A contrastive study of the
realization and perception of refusals. xiv, 190 pp. + index. Expected January 2008
170 Hougaard, Anders and Todd Oakley (eds.): Mental Spaces in Discourse and Interaction. Expected
January 2008
169 Connor, Ulla, Edwin R. Nagelhout and William V. Rozycki (eds.): Contrastive Rhetoric.
Reaching to intercultural rhetoric. viii, 319 pp. + index. Expected December 2007
168 Proost, Kristel: Conceptual Structure in Lexical Items. The lexicalisation of communication concepts in
English, German and Dutch. xii, 296 pp. + index. Expected December 2007
167 Bousfield, Derek: Impoliteness in Interaction. xiii, 278 pp. + index. Expected November 2007
166 Nakane, Ikuko: Silence in Intercultural Communication. Perceptions and performance.
xii, 233 pp. + index. Expected November 2007
165 Bublitz, Wolfram and Axel Hübler (eds.): Metapragmatics in Use. vi, 290 pp. + index. Expected
December 2007
164 Englebretson, Robert (ed.): Stancetaking in Discourse. Subjectivity, evaluation, interaction. 2007.
vii, 323 pp.
163 Lytra, Vally: Play Frames and Social Identities. Contact encounters in a Greek primary school.
xi, 291 pp. + index. Expected November 2007
162 Fetzer, Anita (ed.): Context and Appropriateness. Micro meets macro. 2007. vi, 265 pp.
161 Celle, Agnès and Ruth Huart (eds.): Connectives as Discourse Landmarks. 2007. viii, 212 pp.
160 Fetzer, Anita and Gerda Eva Lauerbach (eds.): Political Discourse in the Media. Cross-cultural
perspectives. 2007. viii, 379 pp.
159 Maynard, Senko K.: Linguistic Creativity in Japanese Discourse. Exploring the multiplicity of self,
perspective, and voice. 2007. xvi, 356 pp.
158 Walker, Terry: Thou and You in Early Modern English Dialogues. Trials, Depositions, and Drama
Comedy. 2007. xx, 339 pp.
157 Crawford Camiciottoli, Belinda: The Language of Business Studies Lectures. A corpus-assisted
analysis. 2007. xvi, 236 pp.
156 Vega Moreno, Rosa E.: Creativity and Convention. The pragmatics of everyday figurative speech. 2007.
xii, 249 pp.
155 Hedberg, Nancy and Ron Zacharski (eds.): The Grammar–Pragmatics Interface. Essays in honor of
Jeanette K. Gundel. 2007. viii, 345 pp.
154 Hübler, Axel: The Nonverbal Shift in Early Modern English Conversation. 2007. x, 281 pp.
153 Arnovick, Leslie K.: Written Reliquaries. The resonance of orality in medieval English texts. 2006.
xii, 292 pp.
152 Warren, Martin: Features of Naturalness in Conversation. 2006. x, 272 pp.
151 Suzuki, Satoko (ed.): Emotive Communication in Japanese. 2006. x, 234 pp.
150 Busse, Beatrix: Vocative Constructions in the Language of Shakespeare. 2006. xviii, 525 pp.
149 Locher, Miriam A.: Advice Online. Advice-giving in an American Internet health column. 2006.
xvi, 277 pp.
148 Fløttum, Kjersti, Trine Dahl and Torodd Kinn: Academic Voices. Across languages and disciplines.
2006. x, 309 pp.
147 Hinrichs, Lars: Codeswitching on the Web. English and Jamaican Creole in e-mail communication.
2006. x, 302 pp.
146 Tanskanen, Sanna-Kaisa: Collaborating towards Coherence. Lexical cohesion in English discourse.
2006. ix, 192 pp.
145 Kurhila, Salla: Second Language Interaction. 2006. vii, 257 pp.
144 Bührig, Kristin and Jan D. ten Thije (eds.): Beyond Misunderstanding. Linguistic analyses of
intercultural communication. 2006. vi, 339 pp.
143 Baker, Carolyn, Michael Emmison and Alan Firth (eds.): Calling for Help. Language and social
interaction in telephone helplines. 2005. xviii, 352 pp.
142 Sidnell, Jack: Talk and Practical Epistemology. The social life of knowledge in a Caribbean community.
2005. xvi, 255 pp.
141 Zhu, Yunxia: Written Communication across Cultures. A sociocognitive perspective on business genres.
2005. xviii, 216 pp.
140 Butler, Christopher S., María de los Ángeles Gómez-González and Susana M. Doval-Suárez
(eds.): The Dynamics of Language Use. Functional and contrastive perspectives. 2005. xvi, 413 pp.
139 Lakoff, Robin T. and Sachiko Ide (eds.): Broadening the Horizon of Linguistic Politeness. 2005.
xii, 342 pp.
138 Müller, Simone: Discourse Markers in Native and Non-native English Discourse. 2005. xviii, 290 pp.
137 Morita, Emi: Negotiation of Contingent Talk. The Japanese interactional particles ne and sa. 2005.
xvi, 240 pp.
136 Sassen, Claudia: Linguistic Dimensions of Crisis Talk. Formalising structures in a controlled language.
2005. ix, 230 pp.
135 Archer, Dawn: Questions and Answers in the English Courtroom (1640–1760). A sociopragmatic
analysis. 2005. xiv, 374 pp.
134 Skaffari, Janne, Matti Peikola, Ruth Carroll, Risto Hiltunen and Brita Wårvik (eds.):
Opening Windows on Texts and Discourses of the Past. 2005. x, 418 pp.
133 Marnette, Sophie: Speech and Thought Presentation in French. Concepts and strategies. 2005.
xiv, 379 pp.
132 Onodera, Noriko O.: Japanese Discourse Markers. Synchronic and diachronic discourse analysis. 2004.
xiv, 253 pp.
131 Janoschka, Anja: Web Advertising. New forms of communication on the Internet. 2004. xiv, 230 pp.
130 Halmari, Helena and Tuija Virtanen (eds.): Persuasion Across Genres. A linguistic approach. 2005.
x, 257 pp.
129 Taboada, María Teresa: Building Coherence and Cohesion. Task-oriented dialogue in English and
Spanish. 2004. xvii, 264 pp.
128 Cordella, Marisa: The Dynamic Consultation. A discourse analytical study of doctor–patient
communication. 2004. xvi, 254 pp.
127 Brisard, Frank, Michael Meeuwis and Bart Vandenabeele (eds.): Seduction, Community,
Speech. A Festschrift for Herman Parret. 2004. vi, 202 pp.
126 Wu, Yi’an: Spatial Demonstratives in English and Chinese. Text and Cognition. 2004. xviii, 236 pp.
125 Lerner, Gene H. (ed.): Conversation Analysis. Studies from the first generation. 2004. x, 302 pp.
124 Vine, Bernadette: Getting Things Done at Work. The discourse of power in workplace interaction. 2004.
x, 278 pp.
123 Márquez Reiter, Rosina and María Elena Placencia (eds.): Current Trends in the Pragmatics of
Spanish. 2004. xvi, 383 pp.
122 González, Montserrat: Pragmatic Markers in Oral Narrative. The case of English and Catalan. 2004.
xvi, 410 pp.
121 Fetzer, Anita: Recontextualizing Context. Grammaticality meets appropriateness. 2004. x, 272 pp.
120 Aijmer, Karin and Anna-Brita Stenström (eds.): Discourse Patterns in Spoken and Written Corpora.
2004. viii, 279 pp.
119 Hiltunen, Risto and Janne Skaffari (eds.): Discourse Perspectives on English. Medieval to modern.
2003. viii, 243 pp.
118 Cheng, Winnie: Intercultural Conversation. 2003. xii, 279 pp.
117 Wu, Ruey-Jiuan Regina: Stance in Talk. A conversation analysis of Mandarin final particles. 2004.
xvi, 260 pp.
116 Grant, Colin B. (ed.): Rethinking Communicative Interaction. New interdisciplinary horizons. 2003.
viii, 330 pp.
115 Kärkkäinen, Elise: Epistemic Stance in English Conversation. A description of its interactional
functions, with a focus on I think. 2003. xii, 213 pp.
114 Kühnlein, Peter, Hannes Rieser and Henk Zeevat (eds.): Perspectives on Dialogue in the New
Millennium. 2003. xii, 400 pp.
113 Panther, Klaus-Uwe and Linda L. Thornburg (eds.): Metonymy and Pragmatic Inferencing. 2003.
xii, 285 pp.
112 Lenz, Friedrich (ed.): Deictic Conceptualisation of Space, Time and Person. 2003. xiv, 279 pp.
111 Ensink, Titus and Christoph Sauer (eds.): Framing and Perspectivising in Discourse. 2003. viii, 227 pp.
110 Androutsopoulos, Jannis K. and Alexandra Georgakopoulou (eds.): Discourse
Constructions of Youth Identities. 2003. viii, 343 pp.
109 Mayes, Patricia: Language, Social Structure, and Culture. A genre analysis of cooking classes in Japan and
America. 2003. xiv, 228 pp.
108 Barron, Anne: Acquisition in Interlanguage Pragmatics. Learning how to do things with words in a study
abroad context. 2003. xviii, 403 pp.
107 Taavitsainen, Irma and Andreas H. Jucker (eds.): Diachronic Perspectives on Address Term
Systems. 2003. viii, 446 pp.
106 Busse, Ulrich: Linguistic Variation in the Shakespeare Corpus. Morpho-syntactic variability of second
person pronouns. 2002. xiv, 344 pp.
105 Blackwell, Sarah E.: Implicatures in Discourse. The case of Spanish NP anaphora. 2003. xvi, 303 pp.
104 Beeching, Kate: Gender, Politeness and Pragmatic Particles in French. 2002. x, 251 pp.
103 Fetzer, Anita and Christiane Meierkord (eds.): Rethinking Sequentiality. Linguistics meets
conversational interaction. 2002. vi, 300 pp.
102 Leafgren, John: Degrees of Explicitness. Information structure and the packaging of Bulgarian subjects
and objects. 2002. xii, 252 pp.
101 Luke, K. K. and Theodossia-Soula Pavlidou (eds.): Telephone Calls. Unity and diversity in
conversational structure across languages and cultures. 2002. x, 295 pp.
100 Jaszczolt, Katarzyna M. and Ken Turner (eds.): Meaning Through Language Contrast. Volume 2.
2003. viii, 496 pp.
99 Jaszczolt, Katarzyna M. and Ken Turner (eds.): Meaning Through Language Contrast. Volume 1.
2003. xii, 388 pp.
98 Duszak, Anna (ed.): Us and Others. Social identities across languages, discourses and cultures. 2002.
viii, 522 pp.
97 Maynard, Senko K.: Linguistic Emotivity. Centrality of place, the topic-comment dynamic, and an
ideology of pathos in Japanese discourse. 2002. xiv, 481 pp.
96 Haverkate, Henk: The Syntax, Semantics and Pragmatics of Spanish Mood. 2002. vi, 241 pp.
95 Fitzmaurice, Susan M.: The Familiar Letter in Early Modern English. A pragmatic approach. 2002.
viii, 263 pp.
94 McIlvenny, Paul (ed.): Talking Gender and Sexuality. 2002. x, 332 pp.
93 Baron, Bettina and Helga Kotthoff (eds.): Gender in Interaction. Perspectives on femininity and
masculinity in ethnography and discourse. 2002. xxiv, 357 pp.
92 Gardner, Rod: When Listeners Talk. Response tokens and listener stance. 2001. xxii, 281 pp.
91 Gross, Joan: Speaking in Other Voices. An ethnography of Walloon puppet theaters. 2001. xxviii, 341 pp.
90 Kenesei, István and Robert M. Harnish (eds.): Perspectives on Semantics, Pragmatics, and Discourse.
A Festschrift for Ferenc Kiefer. 2001. xxii, 352 pp.
89 Itakura, Hiroko: Conversational Dominance and Gender. A study of Japanese speakers in first and
second language contexts. 2001. xviii, 231 pp.
88 Bayraktaroğlu, Arın and Maria Sifianou (eds.): Linguistic Politeness Across Boundaries. The
case of Greek and Turkish. 2001. xiv, 439 pp.
87 Mushin, Ilana: Evidentiality and Epistemological Stance. Narrative Retelling. 2001. xviii, 244 pp.
86 Ifantidou, Elly: Evidentials and Relevance. 2001. xii, 225 pp.
85 Collins, Daniel E.: Reanimated Voices. Speech reporting in a historical-pragmatic perspective. 2001.
xx, 384 pp.
84 Andersen, Gisle: Pragmatic Markers and Sociolinguistic Variation. A relevance-theoretic approach to
the language of adolescents. 2001. ix, 352 pp.
83 Márquez Reiter, Rosina: Linguistic Politeness in Britain and Uruguay. A contrastive study of requests
and apologies. 2000. xviii, 225 pp.
82 Khalil, Esam N.: Grounding in English and Arabic News Discourse. 2000. x, 274 pp.
81 Di Luzio, Aldo, Susanne Günthner and Franca Orletti (eds.): Culture in Communication.
Analyses of intercultural situations. 2001. xvi, 341 pp.
80 Ungerer, Friedrich (ed.): English Media Texts – Past and Present. Language and textual structure. 2000.
xiv, 286 pp.
79 Andersen, Gisle and Thorstein Fretheim (eds.): Pragmatic Markers and Propositional Attitude.
2000. viii, 273 pp.
78 Sell, Roger D.: Literature as Communication. The foundations of mediating criticism. 2000. xiv, 348 pp.
77 Vanderveken, Daniel and Susumu Kubo (eds.): Essays in Speech Act Theory. 2002. vi, 328 pp.
76 Matsui, Tomoko: Bridging and Relevance. 2000. xii, 251 pp.
75 Pilkington, Adrian: Poetic Effects. A relevance theory perspective. 2000. xiv, 214 pp.
74 Trosborg, Anna (ed.): Analysing Professional Genres. 2000. xvi, 256 pp.
73 Hester, Stephen K. and David Francis (eds.): Local Educational Order. Ethnomethodological studies
of knowledge in action. 2000. viii, 326 pp.
72 Marmaridou, Sophia S.A.: Pragmatic Meaning and Cognition. 2000. xii, 322 pp.
71 Gómez-González, María de los Ángeles: The Theme–Topic Interface. Evidence from English. 2001.
xxiv, 438 pp.
70 Sorjonen, Marja-Leena: Responding in Conversation. A study of response particles in Finnish. 2001.
x, 330 pp.
69 Noh, Eun-Ju: Metarepresentation. A relevance-theory approach. 2000. xii, 242 pp.
68 Arnovick, Leslie K.: Diachronic Pragmatics. Seven case studies in English illocutionary development.
2000. xii, 196 pp.
67 Taavitsainen, Irma, Gunnel Melchers and Päivi Pahta (eds.): Writing in Nonstandard English.
2000. viii, 404 pp.
66 Jucker, Andreas H., Gerd Fritz and Franz Lebsanft (eds.): Historical Dialogue Analysis. 1999.
viii, 478 pp.
65 Cooren, François: The Organizing Property of Communication. 2000. xvi, 272 pp.
64 Svennevig, Jan: Getting Acquainted in Conversation. A study of initial interactions. 2000. x, 384 pp.
63 Bublitz, Wolfram, Uta Lenk and Eija Ventola (eds.): Coherence in Spoken and Written Discourse.
How to create it and how to describe it. Selected papers from the International Workshop on Coherence,
Augsburg, 24-27 April 1997. 1999. xiv, 300 pp.
62 Tzanne, Angeliki: Talking at Cross-Purposes. The dynamics of miscommunication. 2000. xiv, 263 pp.
61 Mills, Margaret H. (ed.): Slavic Gender Linguistics. 1999. xviii, 251 pp.
60 Jacobs, Geert: Preformulating the News. An analysis of the metapragmatics of press releases. 1999.
xviii, 428 pp.
59 Kamio, Akio and Ken-ichi Takami (eds.): Function and Structure. In honor of Susumu Kuno. 1999.
x, 398 pp.
58 Rouchota, Villy and Andreas H. Jucker (eds.): Current Issues in Relevance Theory. 1998. xii, 368 pp.
57 Jucker, Andreas H. and Yael Ziv (eds.): Discourse Markers. Descriptions and theory. 1998. x, 363 pp.
56 Tanaka, Hiroko: Turn-Taking in Japanese Conversation. A Study in Grammar and Interaction. 2000.
xiv, 242 pp.
55 Allwood, Jens and Peter Gärdenfors (eds.): Cognitive Semantics. Meaning and cognition. 1999.
x, 201 pp.
54 Hyland, Ken: Hedging in Scientific Research Articles. 1998. x, 308 pp.
53 Mosegaard Hansen, Maj-Britt: The Function of Discourse Particles. A study with special reference
to spoken standard French. 1998. xii, 418 pp.
52 Gillis, Steven and Annick De Houwer (eds.): The Acquisition of Dutch. With a Preface by Catherine
E. Snow. 1998. xvi, 444 pp.
51 Boulima, Jamila: Negotiated Interaction in Target Language Classroom Discourse. 1999. xiv, 338 pp.
50 Grenoble, Lenore A.: Deixis and Information Packaging in Russian Discourse. 1998. xviii, 338 pp.
49 Kurzon, Dennis: Discourse of Silence. 1998. vi, 162 pp.
48 Kamio, Akio: Territory of Information. 1997. xiv, 227 pp.
47 Chesterman, Andrew: Contrastive Functional Analysis. 1998. viii, 230 pp.
46 Georgakopoulou, Alexandra: Narrative Performances. A study of Modern Greek storytelling. 1997.
xvii, 282 pp.
45 Paltridge, Brian: Genre, Frames and Writing in Research Settings. 1997. x, 192 pp.
44 Bargiela-Chiappini, Francesca and Sandra J. Harris: Managing Language. The discourse of
corporate meetings. 1997. ix, 295 pp.
43 Janssen, Theo and Wim van der Wurff (eds.): Reported Speech. Forms and functions of the verb.
1996. x, 312 pp.
42 Kotthoff, Helga and Ruth Wodak (eds.): Communicating Gender in Context. 1997. xxvi, 424 pp.
41 Ventola, Eija and Anna Mauranen (eds.): Academic Writing. Intercultural and textual issues. 1996.
xiv, 258 pp.
40 Diamond, Julie: Status and Power in Verbal Interaction. A study of discourse in a close-knit social
network. 1996. viii, 184 pp.
39 Herring, Susan C. (ed.): Computer-Mediated Communication. Linguistic, social, and cross-cultural
perspectives. 1996. viii, 326 pp.
38 Fretheim, Thorstein and Jeanette K. Gundel (eds.): Reference and Referent Accessibility. 1996.
xii, 312 pp.
37 Carston, Robyn and Seiji Uchida (eds.): Relevance Theory. Applications and implications. 1998.
x, 300 pp.
36 Chilton, Paul, Mikhail V. Ilyin and Jacob L. Mey (eds.): Political Discourse in Transition in Europe
1989–1991. 1998. xi, 272 pp.
35 Jucker, Andreas H. (ed.): Historical Pragmatics. Pragmatic developments in the history of English. 1995.
xvi, 624 pp.
34 Barbe, Katharina: Irony in Context. 1995. x, 208 pp.
33 Goossens, Louis, Paul Pauwels, Brygida Rudzka-Ostyn, Anne-Marie Simon-
Vandenbergen and Johan Vanparys: By Word of Mouth. Metaphor, metonymy and linguistic action
in a cognitive perspective. 1995. xii, 254 pp.
32 Shibatani, Masayoshi and Sandra A. Thompson (eds.): Essays in Semantics and Pragmatics. In
honor of Charles J. Fillmore. 1996. x, 322 pp.
31 Wildgen, Wolfgang: Process, Image, and Meaning. A realistic model of the meaning of sentences and
narrative texts. 1994. xii, 281 pp.
30 Wortham, Stanton E.F.: Acting Out Participant Examples in the Classroom. 1994. xiv, 178 pp.
29 Barsky, Robert F.: Constructing a Productive Other. Discourse theory and the Convention refugee
hearing. 1994. x, 272 pp.
28 Van de Walle, Lieve: Pragmatics and Classical Sanskrit. A pilot study in linguistic politeness. 1993.
xii, 454 pp.
27 Suter, Hans-Jürg: The Wedding Report. A prototypical approach to the study of traditional text types.
1993. xii, 314 pp.
26 Stygall, Gail: Trial Language. Differential discourse processing and discursive formation. 1994.
xii, 226 pp.
25 Couper-Kuhlen, Elizabeth: English Speech Rhythm. Form and function in everyday verbal
interaction. 1993. x, 346 pp.
24 Maynard, Senko K.: Discourse Modality. Subjectivity, Emotion and Voice in the Japanese Language.
1993. x, 315 pp.
23 Fortescue, Michael, Peter Harder and Lars Kristoffersen (eds.): Layered Structure and
Reference in a Functional Perspective. Papers from the Functional Grammar Conference, Copenhagen,
1990. 1992. xiii, 444 pp.
22 Auer, Peter and Aldo Di Luzio (eds.): The Contextualization of Language. 1992. xvi, 402 pp.
21 Searle, John R., Herman Parret and Jef Verschueren: (On) Searle on Conversation. Compiled
and introduced by Herman Parret and Jef Verschueren. 1992. vi, 154 pp.
20 Nuyts, Jan: Aspects of a Cognitive-Pragmatic Theory of Language. On cognition, functionalism, and
grammar. 1991. xii, 399 pp.
19 Baker, Carolyn and Allan Luke (eds.): Towards a Critical Sociology of Reading Pedagogy. Papers of the
XII World Congress on Reading. 1991. xxi, 287 pp.
18 Johnstone, Barbara: Repetition in Arabic Discourse. Paradigms, syntagms and the ecology of language.
1991. viii, 130 pp.
17 Piéraut-Le Bonniec, Gilberte and Marlene Dolitsky (eds.): Language Bases ... Discourse Bases.
Some aspects of contemporary French-language psycholinguistics research. 1991. vi, 342 pp.
16 Mann, William C. and Sandra A. Thompson (eds.): Discourse Description. Diverse linguistic analyses
of a fund-raising text. 1992. xiii, 409 pp.
15 Komter, Martha L.: Conflict and Cooperation in Job Interviews. A study of talks, tasks and ideas. 1991.
viii, 252 pp.
14 Schwartz, Ursula V.: Young Children's Dyadic Pretend Play. A communication analysis of plot structure
and plot generative strategies. 1991. vi, 151 pp.

Stancetaking in Discourse Subjectivity Evaluation Interaction PDF

Uploaded by

Copyright:

Available Formats

Stancetaking in Discourse Subjectivity Evaluation Interaction PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Stancetaking in Discourse Subjectivity Evaluation Interaction PDF

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What topics are covered in the book?

What topics are covered in the book?

Stancetaking in Discourse

Pragmatics & Beyond New Series (P&BNS)

John Benjamins Publishing Company

American National Standard for Information Sciences – Permanence of

Library of Congress Cataloging-in-Publication Data

Stancetaking in discourse : subjectivity, evaluation, interaction / edited by Robert

© 2007 – John Benjamins B.V.

Stancetaking in discourse: An introduction 1

Using a corpus to investigate stance quantitatively and qualitatively 27

Linking identity and dialect through stancetaking 49

Grammatical resources for social purposes: Some aspects 

Subjective and intersubjective uses of generalizations 

The stance triangle 139

The role of I guess in conversational stancetaking 183

Stance markers in spoken Finnish: Minun mielestä 

Stancetaking as an interactional activity: Challenging the prior speaker 253

Positioning and alignment as activities of stancetaking in news interviews 283

Name index 319

and performative nature of language by addressing stancetaking from a range of

of human conduct, while highlighting the situated, pragmatic, and interactional

A metalinguistic analysis of stance in everyday language is crucial in com-

2.1 Quantitative findings

This section addresses the quantitative distribution of stance in both corpora. We

Table 1. Tokens of stance in the SBCSAE and BNC

2.2 Qualitative analyses

(1) (Throw Me SBC057: 948.362-966.593)

(2) (Hold my Breath SBC035: 285.730-303.681)

7 I expected to go in there= and talk about,

(3) (Hold my Breath SBC035: 328.955-365.521)

16 what I’m [saying is],

2.3 Collocational evidence

3. Research on stance: Subjectivity, evaluation, interaction

As suggested by the subtitle of the present volume – Subjectivity, evaluation, inter-

Subjectivity, broadly construed, is seen as an essential quality of language. As Ben-

Thus, evaluation can be roughly summed up as subjectivity with a focus. In oth-

Researchers in both corpus linguistics and systemic-functional linguistics have

combinations of subjects, verb types, and tenses in American English, arguing

Scholars working in this approach have focused on analyzing and comparing a

Yet, for researchers whose work is either based in or informed by conversation

Thus, any interactional sequence (such as adjacency pairs, assessment sequences,

alternation) and illustrates that in addition to the traditional functions ascribed to

6. According to the BNC FAQ at http://www.natcorp.ox.ac.uk/faq.xml. ID=corpus most of

Kärkkäinen, E. 2003. Epistemic Stance in English Conversation: A Description of Its Interactional

Since their development in the 1960s, electronic collections of large numbers of

2. An example: Stance adverbials

• stance marking is particularly frequent in conversation;

3. Pattern and stance

Table 1. Meaning group frequency (adapted from Charles 2004: 118)

Before continuing to discuss the problems associated with quantitative methods,

4. The problems of quantification

(9) …this is a fine, scholarly, and relatively short book….

(11) This book features the normal scholarly apparatus of a bibliography…

(14) …this volume comprises a detailed work of original research embellished

but not necessarily so:

5. Corpora and the qualitative investigation of stance

Stancetaking in discourse: An introduction 1

Using a corpus to investigate stance quantitatively and qualitatively 27

Linking identity and dialect through stancetaking 49

Grammatical resources for social purposes: Some aspects

Subjective and intersubjective uses of generalizations

The stance triangle 139

The role of I guess in conversational stancetaking 183

Stance markers in spoken Finnish: Minun mielestä

Stancetaking as an interactional activity: Challenging the prior speaker 253

Positioning and alignment as activities of stancetaking in news interviews 283

Name index 319