Emergent Syntax For Conversation - Clausal Patterns and The Organization of Action

Emergent Syntax
for Conversation
Clausal patterns and the organization
of action
Studies in Language and Social Interaction
Edited by
Yael Maschler, Simona Pekarek Doehler,
Jan Lindström and Leelo Keevallik
32
John Benjamins Publishing Company

Emergent Syntax for Conversation
Studies in Language and Social Interaction (SLSI)
issn 1879-3983
Studies in Language and Social Interaction is a series which continues the tradition of
Studies in Discourse and Grammar, but with a new focus. It aims to provide a forum
for research on grammar, understood broadly, in its natural home environment,
spoken interaction. The assumption underlying the series is that the study of
language as it is actually used in social interaction provides the foundation for
understanding how the patterns and regularities we think of as grammar emerge
from everyday communicative needs. The editors welcome language-related research
from a range of different methodological traditions, including conversation analysis,
interactional linguistics, and discourse-functional linguistics.
For an overview of all books published in this series, please see
benjamins.com/catalog/slsi
Editors
Sandra A. Thompson Elizabeth Couper-Kuhlen
University of California, Santa Barbara, USA University of Helsinki, Finland
Editorial Board
Peter Auer Barbara A. Fox
University of Freiburg, Germany University of Colorado, USA
Galina Bolden Makoto Hayashi
Rutgers University, USA Nagoya University, Japan
Arnulf Deppermann Marja-Liisa Helasvuo
Institut für Deutsche Sprache, Germany University of Turku, Finland
Paul Drew K.K. Luke
University of York, UK Nanyang Technological University, Singapore
Volume 32
Emergent Syntax for Conversation. Clausal patterns and the organization of action
Edited by Yael Maschler, Simona Pekarek Doehler, Jan Lindström
and Leelo Keevallik
Emergent Syntax
for Conversation
Clausal patterns and the organization of action
Edited by
Yael Maschler
University of Haifa
Simona Pekarek Doehler

University of Neuchâtel
Jan Lindström
University of Helsinki
Leelo Keevallik
Linköping University

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

of Paper for Printed Library Materials, ansi z39.48-1984.
doi 10.1075/slsi.32
Cataloging-in-Publication Data available from Library of Congress:
lccn 2019032415 (print) / 2019032416 (e-book)
isbn 978 90 272 0443 1 (Hb)
isbn 978 90 272 6193 9 (e-book)
© 2020 – 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 Company · www.benjamins.com
Table of contents
Chapter 1
Complex syntax-in-interaction: Emergent and emerging clause-
combining patterns for organizing social actions 1
Simona Pekarek Doehler, Yael Maschler, Leelo Keevallik and
Jan Lindström
Part I. Emerging projecting constructions
Chapter 2
Nel senso (che) in Italian conversation: Turn-taking, turn-maintaining and
turn-yielding 25
Elwys De Stefani
Chapter 3
The emergence and routinization of complex syntactic patterns formed
with ajatella ‘think’ and tietää ‘know’ in Finnish talk-in-interaction 55
Ritva Laury and Marja-Liisa Helasvuo
Chapter 4
The insubordinate – subordinate continuum: Prosody, embodied action,
and the emergence of Hebrew complex syntax 87
Yael Maschler
Chapter 5
Emergent patterns of predicative clauses in spoken Hebrew discourse:
The ha'emet (hi) she- ‘the truth (is) that’ construction 127
Hilla Polak-Yitzhaki
Chapter 6
From matrix clause to turn expansion: The emergence of wo juede ‘I feel/
think’ in Mandarin conversational interaction 151
Wei Wang and Hongyin Tao
vi Emergent Syntax for Conversation
Part II. Locally emergent clause-combining patterns
Chapter 7
Practices of clause-combining: From complex wenn-constructions to
insubordinate (‘stand-alone’) conditionals in everyday spoken German 185
Susanne Günthner
Chapter 8
Grammatical coordination of embodied action: The Estonian ja ‘and’ as a
temporal organizer of Pilates moves 221
Leelo Keevallik
Chapter 9
Consecutive clause combinations in instructing activities: Directives and
accounts in the context of physical training 245
Jan Lindström, Camilla Lindholm, Inga-Lill Grahn and
Martina Huhtamäki
Chapter 10
Right-dislocated complement clauses in German talk-in-interaction:
(Re-)specifying propositional referents of the demonstrative pronoun das 275
Nadine Proske and Arnulf Deppermann
Chapter 11
Relative-clause increments and the management of reference:
A multimodal analysis of French talk-in-interaction 303
Ioana-Maria Stoenica and Simona Pekarek Doehler
Chapter 12
Afterword 331
Paul J. Hopper
Index 339
Chapter 1
Complex syntax-in-interaction
Emergent and emerging clause-combining patterns
for organizing social actions
Simona Pekarek Doehler, Yael Maschler, Leelo Keevallik and

Jan Lindström
Speakers do not possess a bird’s eye view of an utterance, but rather move forward
in time through it. (Paul Hopper, 2011: p. 23)
1. Introduction
The past two decades have witnessed a sea-change in our understanding of lan-
guage. Grammar is no longer dominantly seen from a “bird’s eye view” (cf. Hopper,
2011) as an autonomously structured inventory of items and abstract combination
rules, but is increasingly understood as a usage-based, temporal, and ever-adaptive
resource for people’s acting in the social world (Hopper, 1987, 2011; Hakulinen,
2001; Thompson, 2002; Tomasello, 2003; Ellis & Larsen Freeman, 2006; Linell,
2009; Auer, 2009; Bybee, 2010; Fox & Thompson, 2010). The present collection of
original chapters taps into this understanding of language and explores the ways
by which patterns of complex syntax – that is, syntactic structures beyond a simple
clause – relate to the local contingencies of action formation in social interaction,
and how they are tied to participants’ nonverbal (prosodic and/or embodied) con-
duct. The collection investigates both emergent and emerging aspects of grammar
(see the discussions in Hopper, 2011 and Auer & Pfänder, 2011a): it tracks on-line
emergent clause-combining patterns as they are ‘patched together’ on the fly in
response to local interactional contingencies (such as lack of recipient response);
it also investigates emerging grammatical patterns, i.e., patterns that routinize (or:
sediment) in the grammar as interactional resources, for instance for the purpose
of projecting what comes next. We thus focus both on the process of the structur-
ing of patterns of language use in real time and on the results of repeated language
https://doi.org/10.1075/slsi.32.01doe
© 2020 John Benjamins Publishing Company
2 Simona Pekarek Doehler et al.
use in and for social interaction over time, in an attempt to shed light on two facets
of grammar as a highly adaptive resource for interaction.
For the past five decades, scholars working on the social dynamics of con-
versation have seen conversationalists’ use of language as one of the central foci
of analysis. This has resulted in a collaboration with linguists towards “a syntax-
for-conversation”, a concept famously coined by Schegloff (1979). However, the
path towards a micro-socially attuned grammar, which puts the sequential orga-
nization of conversational talk in the foreground, has not been straightforward;
it underwent significant development only rather recently, since the turn of the
21st century, not least through Schegloff ’s visionary paper on the grammar of turn
organization (1996) and the advent of the sub-discipline of interactional linguis-
tics (Selting & Couper-Kuhlen, 2001; Couper-Kuhlen & Selting, 2018; going back
to Ochs, Schegloff & Thompson, 1996). It is in this tradition of interactionally
sensitive research on language structure and the organization of social actions
that we position ourselves, setting a special focus on the centerpiece of traditional
grammatical inquiry, namely, syntax, which we scrutinize in light of its temporal
structuring within situated social interaction.
To date, we have evidence from several languages that the clause – a core
syntactic structure – also represents a relevant unit of interaction (e.g., Helasvuo,
2001; Thompson & Couper-Kuhlen, 2005): Participants orient to clausal struc-
tures for turn-taking, action projection, utterance co-construction, and so on.
Furthermore, both clauses – and more generally grammatical constructions – and
turns at talk have been documented to be configured in real time, moment-by-
moment, during the temporal unfolding of talk-in-interaction, based on close
collaboration between speakers and recipients (Goodwin, 1981; Goodwin &
Goodwin, 1987; Hopper, 1987, 2011; Auer, 1992, 2009). At any moment in time,
a given syntactic trajectory may be revised or expanded beyond its first (second,
etc.) completion point, as illustrated by the phenomenon of increments (Schegloff,
1996; Ford, Fox & Thompson, 2002; Couper-Kuhlen & Ono, 2007). Moreover, at
any moment in time, the trajectory-so-far projects (Auer, 2005) a range of pos-
sible continuations or next actions, which may result in simple or complex clausal
structures within the realm of a speaker’s turn – or (collaboratively) across turns.
This raises a central question: How does the organization of complex syntax in real
time (i.e., in the very process of its production) relate to the on-line unfolding of turns
and actions, and hence to such fundamental tasks in social interaction as action
projection, formation and ascription?
The present volume is designed to address this issue within a variety of
everyday and institutional settings, and across a range of languages: English,
Estonian, Finnish, French, German, Hebrew, Italian, Mandarin, and Swedish. The
authors of each of the chapters set out to uncover how complex syntactic patterns,
Chapter 1. Complex syntax-in-interaction 3
or routinized fragments of what are considered ‘canonical’ patterns, serve as

resources for projecting and linking upcoming turns or actions (cf. e.g., Hopper,
2004; Auer, 2005; Hopper & Thompson, 2008), shedding light on the ways such
patterns emerge incrementally, turn-by-turn, out of the real-time unfolding
trajectories of action (cf. Schegloff 1996; Ford, Fox, and Thompson, 2002). The
chapters in this volume are concerned with the manner by which formal variants
of different clause-combining patterns relate to different social actions, examining
how speakers employ specific patterns of clause combination in order to accom-
plish specific actions, and how these patterns are oriented to by co-participants.
Much attention is paid to the ways all of the above play out within the multimodal
ecologies of interaction, involving grammatical, prosodic and embodied resources
of meaning-making.
In a nutshell, the papers in this volume explore the ways in which on-line
revisions of clause-combining patterns or expansions of these, as well as their
routinization, relate to the building of social actions in real time. In doing so, they
offer a radical alternative to a traditional understanding of clause-combining and
complex syntax as resulting from formal combination rules.
2. Complex syntax-in-interaction
As a consequence of this new conceptualization of grammar as a locally sensi-

tive, temporally unfolding resource for social interaction, linguistic structures are
increasingly studied in their natural ecology, i.e., talk-in-interaction, in relation
to other semiotic resources, such as prosody, gaze, gesture, and the moving body
(e.g. Hayashi, 2005; Keevallik, 2013, 2018; Mondada, 2014a, Pekarek Doehler,
2019). Owing to a robust literature in interactional linguistics in particular, we
are beginning to understand how specific linguistic constructions function as
interactional resources for accomplishing specific social actions in a growing
number of languages.
We also have evidence for the fact that grammatical constructions are not used
as ready-made templates, but are adapted in the very course of their production
in response to local interactional contingencies such as co-participants’ conduct
so that syntactic trajectories can be re-oriented or expanded in real-time. For ex-
ample, linguistic structures such as increments and left-dislocations may emerge
across interacting parties (Pekarek Doehler et al., 2015; Mondada, 2016), and
variably-routinized segments of syntax come to constitute systematic practices for
projection of talk (Maschler, 2009, 2015; Auer & Maschler, 2013; Keevallik, 2011;
Pekarek Doehler, 2011a, 2015)1 or practices for projection of bodily demonstra-

tions (Streeck, 2002; Keevallik, 2015). These studies show that interactional use
shapes grammar, which is accordingly understood as a continually evolving set of
constructional patterns emerging from recurrent social actions.
The chapters in the present volume set out to explore the formatting of social
actions that involves complex syntactic practices of different structural types,
comprising subordination and coordination in various forms, as well as structur-
ally more ambiguous concatenations that involve juxtaposition or a leaning to-
wards insubordination (cf. Evans, 2007; Evans & Watanabe, 2016; Günthner, 2017;
Maschler, 2018). In particular subordination phenomena have been identified as
being in need of re-thinking in a number of earlier studies on the syntax of spoken
language (e.g. Van Valin, 1984; Thompson, 2002; Lindström & Londen, 2008; Laury
& Suzuki, 2011). For instance, subordinating structures have been considered to
be part of bi-clausal constructions in which one of the clauses (the ‘dependent
clause’) is embedded as one of the arguments or adverbial complements of the
main verb, or modifies one of the nominal constituents. However, once investi-
gated in actual spoken language use, this traditional description of subordination
is often found lacking (for pioneering usage-based statements, see Matthiessen
& Thompson, 1988; Thompson & Mulac, 1991). It has been shown, for example,
that the ‘main clause’ of what is classically considered a bi-clausal pattern (e.g.,
involving a complement-taking-predicate) often occurs in talk-in-interaction as
a routinized “fragment” (e.g., English I dunno), that is, a grammaticized marker
of epistemic/evidential/evaluative stance.2 ‘Main clauses’ have also been shown to
function as projecting constructions, or even as prototypical discourse markers
(Maschler, 2009), that foreshadow certain types of action- or turn-trajectories
to come (Thompson, 2002; Maschler, 2012, 2017; Pekarek Doehler, 2016; Polak-
Yitzhaki & Maschler, 2016; Deppermann, 2011; Keevallik, 2011). For instance,
French je sais pas ‘I don’t know’, in its morphophonologically reduced forms chais
pas or ch’pas, can be used as a device for projecting a dispreferred response, as il-
lustrated in Excerpt 1. In line 01, Marie asks Julie why she chose to learn languages
by means of immersion rather than via standard instruction:
1. See also the papers in Laury & Suzuki, 2011; Auer & Pfänder, 2011b; Deppermann &
Günthner, 2015.
2. See, e.g., Thompson, 2002; Kärkkäinen, 2003 on English; Keevallik, 2006 on Estonian; Imo,
2007 on German; Laury & Okamoto, 2011 and Laury & Ono, 2014 on Finnish and Japanese;
Maschler, 2012, 2017 on Hebrew; and the collection of studies in Lindström et al. 2016b.
(1) (Pekarek Doehler, 2016, Coll_JSP_90)

01 MAR: et- pourquoi.
and why
02 (0.8) ((sliding sound, followed by two short knocks))
03 JUL: be:n >ch’p↑as< je trouve c’est mi↑eux^he.

well dunno I find it’s better
The delay at line 02, the be:n ‘well’-prefacing and the ch’pas ‘dunno’ conspire to project
a dispreferred response: Julie’s je trouve c’est mi↑eux^he ‘I find it’s better’ clearly does
not conform to the terms of Marie’s question, which implicates an account for Julie’s
preferring immersion over standard instruction as a relevant next; Julie, however,
simply re-affirms that she prefers immersion. Further evidence of I-don’t-know-type
of constructions that have routinized as discourse marker-like elements projecting
dispreferred next actions in other languages is provided by Helmer et al. (2016) for
German and by Maschler (2017) and Maschler & Dori-Hacohen (2018) for Hebrew.
Routinization as a projecting construction has also been evidenced, across
several languages, for the initial pieces of bi-clausal patterns of pseudo-cleft
constructions of the type What we need to do is X (e.g., English what we’ll do,
what I’m saying, what happened) and of initial extrapositions such as English it
turns out … or it was really nice … (Hopper, 2001, 2004; Hopper & Thompson,
2008; Couper-Kuhlen & Thompson, 2008 for English; Günthner, 2006; Günthner
& Hopper, 2010 for German; Pekarek Doehler, 2011a, 2015 for French; Maschler
& Fishman, forthcoming for Hebrew). These studies converge on identifying two
key features. First, the initial parts of what are traditionally considered bi-clausal
patterns also occur regularly in talk-in-interaction as syntactically independent
pieces, i.e., stretches of talk that are not syntactically related to subsequent talk.
Second, these pieces function as routinized resources for dealing with a central
exigency of social interaction, namely the need to project what comes next. Both
of these issues are illustrated in Excerpt (2) from a Hebrew conversation between
three young women and a young man:
(2) (Maschler & Fishman, forthcoming)
101 Liraz: ...ze bixlal,
it at_all
it’s not at all,
102 lo ma she-'amarti,
neg what that-say.past.1sg
what I said,
103?(fem): @
104 Liraz: ma she-'amarti,
what that-say.past.1sg
what I said,
105 ...hi lo tsrixa lehagid l-o,

she neg need.f.sg say.inf to-3m.sg
she doesn’t have to say to him,
106 ?fem: @
107 Liraz: takshiv,
listen.fut.2m.sg
listen,
108 'a
109 ...hit'oreru b-i regashot,
stirred_up.past.3pl in-1sg feelings
[new] feelings have stirred up in me,
(continued for 8 more intonation units)
In (2) ma she-'amarti, ‘what I said’ (l. 104) projects a rephrasal of the speaker’s
(Liraz’s) previous talk, in order to clarify it and oppose her interlocutor, who
had misrepresented it in her view. What follows is a stretch of discourse much
longer than a clause that is not marked as being syntactically related to the initial
piece – it lacks both the copula ze ‘is’ and the subordinator she- ‘that’ necessary in a
‘canonical’ Hebrew pseudo-cleft construction (i.e., ma she-'amarti ze she-… ‘what
I said is that ….’). Thus, a fragment of the so-called canonical pseudo-cleft – ma
she-'amarti, ‘what I said’ – has become routinized for the interactional purpose
of projecting, and in this case specifically projecting the rephrasal of one’s prior
utterance (Maschler & Fishman, forthcoming).
Furthermore, it has been shown that a structure that looks like a ‘dependent’
clause may be syntactically more or less free-standing, or ‘insubordinate’ (Evans,
2007; Wide, 2014; Lindström et al., 2016a; Günthner, 2017), or be sequentially
rather than grammatically embedded, and that subordinating conjunctions may
function as discourse markers that structure turn-taking and participation
(Keevallik, 2008; Koivisto et al., 2011; Maschler, 2018). For instance, causal clauses
have been shown to be added post-hoc, incrementally, e.g., in pursuit of recipient
response (Ford, 1993; Couper-Kuhlen, 2012), and conditional clauses to address a
potential social sensitivity of the prior social action (Eriksson, 2001):
(3) (Eriksson, 2001: p. 18)
01 ADA: Hur gammal e du?
how old are you
How old are you?
02 (.)
03 ADA: Om jag får fråga [en så frä-]
if I may ask.inf a such bold-
If I may ask such a bold-
04 BRI: [Åttitre ]
Eighty three
05 ADA: Oj oj oj °jösses°
oh oh oh Christ
Oh dear.
In this Swedish conversation, the conditional clause in line 3 can be argued to

be prompted by the minimal pause before Britta begins to respond, indicating a
potential problem with the question. Thus, the clause-combining pattern canoni-
cally known as [main clause + adverbial clause] emerges on the fly, in response to
this precise local interactional contingency.
Crucially, the real-time production of language regularly results in incremen-
tally emergent grammar, including incrementally emergent complex syntactic
patterns. For instance, speakers have been shown to add relative clauses post-hoc,
as increments, to a preceding clause in the pursuit of recipient alignment (see Clift,
2007 on English; Oloff & Havlik, 2018 on Czech) or in response to the recipient’s
display of trouble (Stoenica, 2014 on French). This phenomenon is illustrated here
with the following excerpt from a Finnish telephone conversation, where Vikke
tells her friend Missu which mutual acquaintances she had seen in a dream.
(4) (Laury & Helasvuo, 2015: p. 160–161)
01 Vikke: No siel oli, (.) khm (0.3)
prt dem+loc be.pst+3sg
Well there was
02 .mt mun ↑serkkuni <Salla>,
1sg.gen cousin.1sgpx Name.fem
my cousin Salla,
03 Missu: Nii.
prt
Yeah.
04 Vikke: Rantalan Anni,
LastName.gen Name.fem
Anni Rantala.
05 Missu: £Joo-o£?,
prt
Yeah?
06 (.)
07 Vikke: ja sit se yks poika: Lahdesta.
and then dem3 one guy PlaceName.ela
and then that one guy from Lahti.
08 (.)
09 Mikä on, (0.6) oli, v- vaihtooppilaana.h
rel be+3sg be.pst+3sg exchange.student.ess
Who is, was, away as an exchange student
10 (0.3)
11 Vikke: just vuaden?, se, Mäkelän Kimmon kaveri.

just year.acc dem3 LastName.gen name.masc.gen friend
for one year?, that friend of Kimmo Mäkelä’s.
12 (0.3)
13 Missu: KUka,h
who
WHO,
14 Vikke: Se mikä on sen, (0.3) sen,
dem3 rel be+3sg dem3.gen dem3.gen
The one who is that, that,
15 sen poliisim [poika.
dem3.gen police.gen son
that policeman’s son.
16 Missu: [.hhh Se Vehviläine,h
dem3 LastName
That Vehviläinen.
17 Vikke: >Nii<.
prt
Right.
The first two persons mentioned in the list (l. 2 and 4) are unproblematic, but
identification becomes an issue with the third item, se yks poika: Lahdesta ‘that
one guy from Lahti’ (l. 7). This prosodically complete turn part is followed by
a micro pause in line 8, after which the speaker adds a relative clause, thereby
enhancing the identifiability of the person in question (l. 9). Some trouble still
ensues, until the recipient Missu, after some further clues from Vikke, displays
recognition in line 16.
More generally, these kinds of syntactic expansions have also been docu-
mented to involve revisions of syntactic trajectories so-far, often entailing change
in constituent status in pivot-like constructions (Horlacher & Pekarek Doehler,
2014; the collection of papers in Norén & Linell, 2013; Pekarek Doehler, 2011b;
Walker 2007). Take the following classical example quoted by Walker (2007), in
which Lottie and Emma talk about buying a turkey for Thanksgiving dinner (the
pivot pattern is highlighted in bold):
(5) (Walker, 2007: p. 2219)
01 Lot: I went down there and got a- (.) Rancho a fresh one
02 (0.6)
03 Emm: oh that’s what I’d like to have is a fresh one
The segment what I’d like to have (l. 3) represents the pivot element forming a
grammatical unit both with what precedes and with what follows: It appears first
as the second part of what is known in the literature as an inverted pseudo-cleft
construction (that’s what I’d like to have), and then as the first part of a pseudo-cleft
construction (what I’d like to have is a fresh one). Phenomena of this kind show
us that the syntactic constituency of the turn-in-progress can be changed on the
fly. Pivot constructions, just like many of the clause-combining patterns discussed
above, are local constructs resulting from the “patching together” (Pekarek Doehler
& Horlacher, 2013) of linguistic resources in real time. A further example of a
“patched-up” type of utterance is provided by Hopper’s (2011) analysis of ‘sluicing’
(Ross, 1969), as in She said okay, I’ll stay another year, but, I don’t know why, where
the boldfaced segment is shown to be an incremental addition of a ready-made
piece, rather than part A of a bi-clausal construction which is considered to be an
elliptical contraction of a longer sentence (such as I don’t know why she said it).
Finally, recent evidence suggests that speakers’ use of grammar is tightly related
to their deployment of embodied resources, such as gaze, body movements and
moving through space (Keevallik, 2013, 2014; Iwasaki, 2011; Broth & Mondada,
2013; Mondada, 2009, 2014a, Pekarek Doehler, 2019). For example, Keevallik
(2013) analyzes speaker turns in dance class interaction in which a syntactically
incomplete clause is followed by a bodily demonstration, such as in the following,
produced by an instructor in a Swedish couple-dance class:
(6) (Keevallik, 2013: p. 8)
01INS För om killarna gör, *(2.0)*
because if guy.pl.def do
Because if the guys do
*step step*
02 Så kommer ju tjejerna att göra, *(1.3)*
then fut prt lady.pl.def inf do.inf
then the ladies will do
*step step*
03 eftersom de följer. Eller hur.
because they follow. Right?
This turn is composed of two syntactically incomplete clauses, each one

projecting a bodily demonstration. Together, the two tokens of [incomplete
clause+demonstration] form a complex clausal structure of the if … then type.
Keevallik argues that the classic list of unit-types comprising turn-constructional
units (Sacks, Schegloff, and Jefferson, 1974) should be expanded to include
syntactic-bodily constructions. In the same vein, our conceptions of syntactic
patterns of clause-combining could be expanded by considering syntactic-bodily
constructions. Multimodal research shows how embodied conduct plays into such
fundamental issues as turn-taking (see pioneering work by Goodwin & Goodwin,
1987; Goodwin, 2000; Mondada, 2007) and projection of emergent structures
though gaze and gesture (Mondada, 2006; Streeck & Jordan, 2009). Importantly,
it has been demonstrated that the real-time production of an utterance may be
affected by others’ embodied conduct (see e.g. Goodwin, 1981; Mondada, 2009;
papers in Streeck, Goodwin, & Le Baron, 2011; Sorjonen & Raevaara, 2014), and
that such conduct may in turn trigger on-line emergent clause-combining pat-
terns (Stoenica, 2018). Such evidence calls for work carrying us further toward
a more holistic understanding of the functioning of language and grammar as
situated in the multisemiotic ecologies of social interaction (Mondada, 2014b,
2016, Keevallik, 2018).
3. Purpose and structure of the volume
The present volume contributes original empirical work to the above line of inves-
tigation. It offers a collection of chapters that study complex syntax in interaction
and provide new findings across a range of languages stemming from different
(branches of) language families: Finnish and Estonian (Finno-Ugric), French,
Italian, German, and Swedish (from the Romance and Germanic branches of
Indo-European), Hebrew (Semitic) and Mandarin (Sino-Tibetan). The volume as
a whole expands our understanding of the functioning-in-use of natural languages
through the scrutiny of a basic over-arching syntactic operation – clause-combin-
ing – and provides support for a conception of linguistic units as fundamentally
temporal, emergent, and sensitive to local interactional contingencies.
This introduction is followed by ten chapters presenting empirical research
on clause-combining practices as they emerge in the temporal, interactional, and/
or embodied context of talking. The research chapters are organized according
to the ways in which they address two key themes: (i) emerging projecting con-
structions, and (ii) locally-emergent clause-combining patterns. As both themes
can be intertwined with one another, the chapters naturally also cross these
thematic boundaries.
i. Emerging projecting constructions
Focusing attention on linguistic forms that have to do with the formation of com-
plex syntactic structures, the chapters in this section are concerned with gram-
matical, sequential, and prosodic aspects of sedimentation, most prominently in its
relation to the interactional work of projection. These chapters show in particular
how complex syntactic patterns (or pieces of ‘canonical’ types of such patterns)
routinize for the purpose of projecting upcoming structural configurations and
trajectories of actions, or relating ongoing actions to prior turns and actions.
The chapter by De Stefani shows the ways by which the Italian phrase nel senso
(che) ‘in the sense (that)’ is employed as a projector phrase and a connector in
conversation. The phrase, which according to traditional understanding projects

a restrictive relative clause, is regularly used between clauses, the second of which
is not grammatically tied to the first but is pragmatically related to it, specify-
ing, elaborating, or expanding on its content. Nel senso (che) can occur in turn-
beginnings, where it displays the manner by which the incipient turn relates to a
preceding turn. As a device for constructing multi-unit turns, it can be used for
introducing a turn extension at moments of insufficient recipient reaction, such as
lack of verbal or embodied acknowledgement. This study also demonstrates that
the shorter variant of the phrase, nel senso, is deployed turn-finally together with an
open-hand gesture and gaze movement, in order to seek support for the speaker’s
standpoint in a multiparty conversation. Together, the findings provide evidence
for the routinization of the construction for interactional purposes, and show how
these purposes may involve precise embodied conduct on the part of the speaker.
The chapter by Laury and Helasvuo demonstrates more or less fixed and more
or less syntactically complex patterns that occur with two Finnish verbs, ajatella
‘think’, and tietää ‘know’. Based on a quantitative analysis of corpora, the authors
show that while both of these verbs tend to occur in the 1st person singular form,
they differ in their predominant use of tense, polarity, syntax, and projection of ac-
tion. While ajatella is rarely reduced, frequently combined with a complementizer,
only rarely negated, and projects information about the speaker’s own thoughts
and plans, as well as expresses epistemic and evaluative/affective stance, tietää
occurs most regularly in the negated form, often comes entirely without comple-
ments and occurs in response to the prior turn, often to questions, but also to
compliments and other types of first pair parts requiring responses. Tietää is in
itself a projective epistemic expression. The paper furthermore details the pho-
netic reduction of the different structures and ultimately shows the ways by which
structures which would conventionally be analyzed as main clauses are emerging
as formulaic fragments, often with reduced syntactic complexity and considerable
morpho-phonological reduction.
Based on data from Hebrew, Maschler discusses the insubordinate-subordi-
nate continuum of clausal combinations. She examines variations of patterns in
which syntactically integrated and unintegrated she-‘that/which/who’-clauses are
produced following final and continuing intonation contours. A comparison of
patterns in the data shows that the conjunction she- is generally utilized as an ele-
ment that ties back to a preceding stretch of interaction and projects an elaboration
or evaluation of it. This elaboration can result in morpho-syntactically complex
constructions involving relative, complement, or adverbial clauses, but in insub-
ordinate cases the she-clause can also stand on its own. Moreover, it is shown that
elaboration can target not only NPs and clauses, but also gestures. The patterning
of the insubordinate-subordinate continuum suggests that canonical, syntactically
integrated varieties of Hebrew relative, complement, and adverbial clauses stand

out as sedimentations of syntactically less integrated clause combinations that are
frequently deployed in spoken, interactional language.
Polak-Yitzhaki’s chapter on the Hebrew ha'emet (hi) she- ‘the truth (is) that’
construction reveals the ways in which the initial part of a nominal clause can
function as a projective item (perhaps a discourse marker), and consequently,
how it implements specific functions in combination with the ensuing predicative
clause. This chapter offers an account of the ways by which grammatical patterns
constitute routinized solutions for specific communicative needs, which in turn
addresses the central issue of the nature of language. Rather than thinking of
grammar as a generic algorithm capable of expressing every possible proposition,
this study shows how speakers have a range of complex but pragmatically specific
form-meaning packages at their disposal. Speakers employ the focus construction
to formulate a revelation of sensitive information concerning themselves, to set
the record straight regarding some potential misunderstanding of what they have
just said, or to express speaker stance. As an epistemic claim about access to truth,
the construction projects information that the speaker is certain about, which
explains its collocation with statements about self.
The chapter by Wang and Tao presents a study of a Mandarin Chinese mental
verb construction – wo juede ‘I feel/think’ – showing that it has evolved from a ma-
trix clause involving a complement-taking predicate to a projecting construction
introducing a multi-unit turn. The wo juede construction has developed not only
epistemic and evaluative uses in the (inter)subjective domain, as has been shown
for constructions of this type in a variety of languages; it has then developed also a
conversation organizational use as a ‘prospective indexical’ (Goodwin, 1996) that
facilitates the extension of turns – a pattern described for other languages but not
yet for any Sino-Tibetan language. The authors show that this latter use contrasts
with the two earlier uses in terms of both syntax and prosody. Finally, they suggest
an interaction-based explanation for this particular grammaticization path: evalu-
ation sequences often get much extended due to the interactive nature of stance
negotiation among participants. This may cause the link between the matrix clause
and the ensuing complement clauses to weaken, with the result that the matrix
clause gradually drifts to an independent status and exhibits greater dissociation
from the original clause-combining pattern.
ii. Locally emergent clause-combining patterns
The chapters in this section analyze the broad spectrum of clause combining
phenomena from subordination to coordination that are configured in real time
as resources for action formation. Such an analytic scope contributes to our
understanding of what kinds of interactional tasks such on-line grammatical pat-

terning can generally solve, while further testifying to the locally sensitive and lo-
cally configured nature of clause and action combining. The authors demonstrate
the ways in which complex syntactic patterns emerge from the sequential concat-
enation of actions in real time and document the type of interactional work such
emergent patterns accomplish in response to precise interactional contingencies.
Günthner presents formal variations of conditional wenn-clauses in German
talk-in-interaction, ranging from canonical, syntactically integrated (and subordi-
nated) uses of prepositioned wenn-clauses to extraposed, non-integrated and free-
standing conditionals. These varying practices of clause-combining are connected
to the speaker’s sensitivity to contingencies in the temporal unfolding of talk, and
deal with issues of projection, expansion and accomplishment of social action.
Sedimented bi-clausal patterns are a powerful resource for projection, interaction-
ally most notably manifest in collaborative productions. The analysis shows that
syntactic and prosodic non-integration of wenn-clauses transforms them into
turn segments that prepare the recipient for a more focused communicative ac-
tion, for example, when staging the punchline of a narrative. Loosely-integrated
conditionals are also used in expanded turns that may incorporate several turn-
constructional units, serving to reserve the floor until a foreshadowed consequence
or conclusion is reached. Finally, wenn-clauses can stand alone, implementing a
social action in their own right, typically in expressing wishes, requests, threats or
exclamative stances, in which the conditional form is a way of dealing with possi-
bly sensitive actions. Such uses call into question the traditional binary categories
of dependent and independent clauses, a problem that the author discusses with
reference to insubordination.
Keevallik’s paper examines the Estonian coordinating conjunction ja ‘and’ as
it is used by instructors in Pilates classes. Ja is employed either for the overall struc-
turing of the class (moving into a next exercise) or for the organization of different
steps within a given exercise (e.g., notifying the students of an upcoming return to
the starting position). The use of ja for exercise-organizational purposes contrasts
with the fact that the coordination of clauses is otherwise accomplished by the
instructor without an explicit coordinator, by mere juxtaposition. Furthermore,
it is shown how ja can be used as a practical means for the temporal extension of
talk to achieve synchronicity with embodied behavior: the instructor extends ja
through syllable lengthening while waiting for the students to arrive at a certain
position for her next utterance to be appropriately timed with the student’s next
move. In Pilates classes, then, the clause-combining pattern of the coordination
type emerges over time in ways that are responsive to co-participants’ bodily
movements and the conjunction functions as a temporal index of the embodied
trajectories of action. Prosodic features are shown to differentiate uses of different
types; for instance, ja produced with a distinct low-fall of pitch contour is used to
provide a starting signal for an exercise. The findings offer compelling evidence for
the ways in which participants’ local meaning-making rests on a complex interplay
between vocal (grammar, prosody, voice quality) and bodily conduct.
The chapter by Lindström, Lindholm, Grahn, and Huhtamäki demonstrates a
generic syntactic pattern consisting of an imperative, declarative, or phrasal first
part followed by a consecutive clause as a second part. The pattern accomplishes a
combination of a directive action and an account for it in the context of individual
physical training. The trainer can produce the instructional pattern as a single
whole, or pause for his/her own demonstration or for compliance by the trainee.
These options occur in different phases of the training process: while trainers’
demonstrations are produced during the preparatory phases, the trainees’ compli-
ance happens during the practice itself. The authors argue that the first part of
the pattern most often prosodically projects the second part, even though that
first part is pragmatically complete. Most importantly, the clause combination
emerges across participants and modalities; it is timed with and locally condi-
tioned by the embodied behavior of the trainer and the trainee. The paper thus
demonstrates that the fact interactants have bodies may be consequential for
grammar. At the same time, it shows that a close scrutiny of function in specific
activities may reveal grammatical regularities that could not have been found in
abstract theorizing, such as an adverb and consecutive conjunctions constituting
a functional paradigm.
The chapter by Proske and Deppermann examines the combination of matrix
clause plus right-dislocated complement clause with the subordinating conjunc-
tion dass ‘that’ in German interactions. In this combination, the first clause contains
the back-linking demonstrative pronoun das ‘this/that’, and the second is a dass-
complement clause that specifies the propositional reference of the demonstrative;
for example, das hab ich nich mitbekommen. (0.32) dass es da so YouTubevideos
gab. ‘I wasn’t aware of that. That there were videos about that on YouTube.’ The
clause-combination, then, consists of an action combination: a first responsive
action followed by a commentary on it that can involve a perspective shift. The au-
thors show that this constructional pattern facilitates an incremental constitution
of meaning and reference, which can be related to responding to another speaker’s
prior turn or tying together parts of the current speaker’s multi-unit turn. They
provide further evidence showing how various bi-clausal constructions (the one
they studied, along with e.g. extrapositions or pseudo-clefts) can be produced as
two separate “chunks”, each of which having distinct interactional functionalities
in everyday language use.
The chapter by Stoenica and Pekarek Doehler is a demonstration of the ways
in which the recipient’s bodily behavior affects the emergence of relative clauses
in French talk-in-interaction. It argues that a clausal combination consisting of a

“main” clause and a “subordinate relative” clause may be produced in response to
local interactional contingencies, such as a puzzled face of the interlocutor, and
that the resulting clause-combination is only retrospectively analyzable as such.
Focusing on self-increments, the study documents the manner in which referen-
tial repair with a relative clause is produced following a vocal or non-vocal display
of trouble by the recipient, while referential elaboration via a relative clause may
ensue following the recipient’s display of recognition of the referent. The paper
also shows that increments, which have typically been analyzed as continuations
of the same action, may accomplish actions of their own. It furthermore chal-
lenges the traditional account of restrictive vs. non-restrictive relative clauses
by showing examples that contradict classical claims (e.g., proper nouns being
followed exclusively by non-restrictive relative clauses). Finally, the paper makes
a compelling case that an incrementally added relative clause can deal with, or
pre-empt, trouble while maximizing progressivity of talk.
Our volume ends with an Afterword written by Paul Hopper, bringing into
sharp relief two threads running through the ten chapters – temporality (“time in-
sofar as it manifests itself in human existence”, Hoy, 2009: p. xiii, cited in Hopper,
this volume), and dialogue – as the two theoretical keystones of interactional
linguistics.
We wish to thank Sandy Thompson and Betty Couper-Kuhlen, the Studies
in Language and Social Interaction series editors, for their critical reading of all
chapters and for their invaluable advice. Thanks are due also to Isja Conen at John
Benjamins for inviting us to embark on this publication, and to Nikolaus Wildner
for his meticulous work on the index.
References
Auer, P. (1992). The neverending sentence: Rightward expansion in spoken language. In M.

Kontra, & T. Váradi (Eds.), Studies in Spoken Language: English, German, Finno-Ugric
(pp. 41–59). Budapest: Hungarian Academy of Science.
Auer, P. (2005). Projection in interaction and projection in grammar. Text 25, 7–36.
https://doi.org/10.1515/text.2005.25.1.7
Auer P. (2009). On-line syntax: Thoughts on the temporality of spoken language. Language Sci-
ences 31, 1–13. https://doi.org/10.1016/j.langsci.2007.10.004
Auer, P., & Pfänder, S. (Eds.). (2011a). Constructions: Emergent or emerging? In P. Auer, & S.
Pfänder (Eds.), Constructions: Emerging and emergent (pp. 2–21). Berlin: de Gruyter.
https://doi.org/10.1515/9783110229080.1
Auer, P., & Pfänder, S. (2011b). Constructions: Emerging and emergent. Berlin: de Gruyter.
https://doi.org/10.1515/9783110229080
Auer, P., & Maschler, Y. (2013). Discourse or Grammar? VS patterns in spoken Hebrew and
spoken German narratives. Language Sciences 37, 147–181.
https://doi.org/10.1016/j.langsci.2012.08.005
Broth, M., & Mondada, L. (2013). Walking away: The embodied achievement of activity closings
in mobile interaction. Journal of Pragmatics 47(1), 41–58.
https://doi.org/10.1016/j.pragma.2012.11.016
Bybee, J. (2010). Language, usage and cognition. Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511750526
Clift, R. (2007). Grammar in time: The non-restrictive ‘which’-clause as an interactional re-
source. Essex Research Reports in Linguistics 55, 51–82.
Couper-Kuhlen, E. (2012). Turn continuation and clause combinations. Discourse Processes
49(3/4), 273–299. https://doi.org/10.1080/0163853X.2012.664111
Couper-Kuhlen, E., & Ono, T. (2007). ‘Incrementing’ in conversation: A comparison of practices
in English, German, and Japanese. Pragmatics 17(4), 513–552.
https://doi.org/10.1075/prag.17.4.02cou
Couper-Kuhlen, E., & Selting, M. (2018). Interactional linguistics: Studying language in social
interaction. Cambridge: Cambridge University Press.
Couper-Kuhlen, E., & Thompson, Sandra A. (2008). On assessing situations and events in
conversation: ‘Extraposition’ and its relatives. Discourse Studies 10(4), 443–467.
https://doi.org/10.1177/1461445608091882
Deppermann, A. (2011). Constructions vs. lexical items as sources of complex meanings: A
comparative study of constructions with German verstehen. In P. Auer, & S. Pfänder (Eds.),
Constructions: Emerging and emergent (pp. 88–126). Berlin: de Gruyter.
https://doi.org/10.1515/9783110229080.88
Deppermann, A., & Günthner, S. (Eds.). (2015). Temporality in interaction. Amsterdam: Ben-
jamins. https://doi.org/10.1075/slsi.27
Ellis, N. C., & Larsen-Freeman, C. (2006). Language emergence: Implications for applied lin-
guistics. Applied Linguistics 27, 558–589. https://doi.org/10.1093/applin/aml028
Evans, N. (2007). Insubordination and its uses. In I. Nikolaeva (Ed.), Finitness: Theoretical and
empirical foundations (pp. 366–431). Oxford: Oxford University Press.
Evans, N., & Watanabe, H. (2016). Insubordination. Amsterdam: Benjamins.
https://doi.org/10.1075/tsl.115
Eriksson, M. (2001). Syntaxens sociala sida [The social aspect of syntax]. Språk och Stil 11, 5–24.
Ford, C. E. (1993). Grammar in interaction: Adverbial clauses in American English conversation.
Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511554278
Ford, C. E., Fox, B. A., & Thompson, S. A. (2002). Constituency and the grammar of turn
increments. In C. Ford, B. Fox, & S. Thompson (Eds.), The language of turn and sequence
(pp. 14–38). Oxford: Oxford University Press.
Fox, B. A., & Thompson, S. A. (2010). Responses to WH-questions in English conversation.
Research on Language and Social Interaction 43(2), 133–156.
https://doi.org/10.1080/08351811003751680
Goodwin, C. (1981). Conversational organization: Interaction between speakers and hearers. New
York: Academic Press.
Goodwin, C. (1996). Transparent vision. In E. Ochs, E. Schegloff, & S. Thompson (Eds.), Interac-
tion and grammar (pp. 370–404). Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511620874.008
Goodwin, C. (2000). Action and embodiment within situated human interaction. Journal of
Pragmatics 32, 1489–1522. https://doi.org/10.1016/S0378-2166(99)00096-X
Goodwin, C., & Goodwin, M. Harness (1987). Concurrent operations on talk: Notes on the
interactive organization of assessments. IPrA Papers in Pragmatics 1(1), 1–55.
https://doi.org/10.1075/iprapip.1.1.01goo
Günthner, S. (2006). “Was ihn trieb, war vor allem Wanderlust” (Hesse: Narziss und Goldmund):
Pseudocleft-Konstruktionen im Deutschen. In S. Günthner, & W. Imo (Eds.), Konstruk-
tionen in der Interaktion (pp. 59–90). Berlin: Mouton de Gruyter.
https://doi.org/10.1515/9783110894158.59
Günthner, S., & Hopper, P. (2010). Zeitlichkeit & sprachliche Strukturen: Pseudo-clefts im Eng-
lischen und Deutschen. Gesprächsforschung – Online-Zeitschrift zur verbalen Interaktion
11, 1–28.
Günthner, S. (2017). Alleinstehende Nebensätze: Insubordinierte wenn-Konstruktionen in der
kommunikativen Praxis. In Y. Ekinci, E. Montanari, & L. Selmani (Eds.): Grammatik und
Variation: Festschrift für Ludger Hoffmann zum 65. Geburtstag (pp. 97–111). Heidelberg:
Synchron Wissenschaftsverlag der Autoren.
Hakulinen, A. (2001). Minimal and non-minimal answers to yes-no questions. Pragmatics
11(1), 1–15. https://doi.org/10.1075/prag.11.1.01hak
Hayashi, M. (2005). Joint turn construction through language and the body: Notes on embodi-
ment in coordinated participation in situated activities. Semiotica, 156, 21–53.
Helasvuo, M.-L. (2001). Syntax in the making: The emergence of syntactic units in Finnish conver-
sational discourse. Amsterdam: Benjamins. https://doi.org/10.1075/sidag.9
Helmer, H., Reineke, S., & Deppermann, A. (2016). A range of uses of negative epistemic con-
structions in German: ICH WEIẞ NICHT as a resource for dispreferred actions. Journal of
Pragmatics 106, 97–114. https://doi.org/10.1016/j.pragma.2016.06.002
Hopper P. J. (1987). Emergent grammar. Berkeley Linguistics Society 13, 139–157.
https://doi.org/10.3765/bls.v13i0.1834
Hopper, P. J. (2001). Grammatical constructions and their discourse origins: Prototype or family
resemblance? In M. Pütz, S. Neimeier, & R. Dirven (Eds.), Applied cognitive linguistics I:
Theory and language acquisition (pp. 109–129). Berlin: Mouton de Gruyter.
https://doi.org/10.1515/9783110866247.109
Hopper, P. J. (2004). The openness of grammatical constructions. Proceedings from the Annual
Meeting of the Chicago Linguistic Society 40(2), 153–175.
Hopper, P. J. (2011). Emergent grammar and temporality in interactional linguistics. In P. Auer,
& S. Pfänder (Eds.), Constructions: Emerging and emergent (pp. 22–45). Berlin: de Gruyter.
https://doi.org/10.1515/9783110229080.22
Hopper, P., & Thompson, S. A. (2008). Projectability and clause combining in interaction. In
R. Laury (Ed.), Crosslinguistic studies of clause combining (pp. 99–123). Amsterdam: Ben-
jamins. https://doi.org/10.1075/tsl.80.06hop
Horlacher, A.-S. & Pekarek Doehler, S. (2014). ‘Pivotage’ in French talk-in-interaction: On the
emergent nature of [clause-NP-clause] pivots. Pragmatics 24(3), 593–622.
https://doi.org/10.1075/prag.24.3.07hor
Hoy, D. C. (2009). The time of our lives: A critical history of temporality. Cambridge, MA: The
MIT Press. https://doi.org/10.7551/mitpress/9780262013048.001.0001
Imo, W. (2007). Construction Grammar und Gesprochene-Sprache-Forschung. Konstruktionen
mit zehn matrixsatzfähigen Verben im gesprochenen Deutsch. Tübingen: Niemeyer.
https://doi.org/10.1515/9783110975895
Iwasaki, S., (2011). The multimodal mechanics of collaborative unit construction in Japanese con-
versation. In J. Streeck, C. Goodwin, & C. LeBaron (eds.), Embodied interaction: Language
and body in the material world (pp. 106–120). Cambridge: Cambridge University Press.
Kärkkäinen, E. (2003). Epistemic stance in English Conversation: A description of interactional
functions, with a focus on I think. Amsterdam: Benjamins.
Keevallik, L. (2006). From discourse pattern to epistemic marker: Estonian (ei) tea ‘don’t know’.
Nordic Journal of Linguistics 29(2), 173–200. https://doi.org/10.1017/S0332586506001570
Keevallik, L. (2008). Conjunction and sequenced actions: The Estonian complementizer and
evidential particle et. In R. Laury (Ed.), Crosslinguistic studies of clause combining: The
multifunctionality of conjunctions (pp. 125–152). Amsterdam: Benjamins.
https://doi.org/10.1075/tsl.80.07kee
Keevallik, L. (2011). Interrogative “complements” and question design in Estonian. In R. Laury,
& R. Suzuki (Eds.), Subordination in conversation: A cross-linguistic perspective (pp. 37–68).
Amsterdam: Benjamins. https://doi.org/10.1075/slsi.24.04kee
Keevallik, L. (2013). The interdependence of bodily demonstrations and clausal syntax. Research
on Language and Social Interaction 46(1), 1–21.
https://doi.org/10.1080/08351813.2013.753710
Keevallik, L. (2014). Turn organization and bodily-vocal demonstrations. Journal of Pragmatics
65, 103–120. https://doi.org/10.1016/j.pragma.2014.01.008
Keevallik, L. (2015). Coordinating the temporalities of talk and dance. In A. Deppermann, & S.
Günthner (Eds.), Temporality in interaction (p. 309–336). Amsterdam: Benjamins.
Keevallik, L. (2018). What does embodied interaction tell us about grammar? Research on
Language and Social Interaction 51, 1–21. https://doi.org/10.1080/08351813.2018.1413887
Koivisto, A., Laury, R., & Seppänen, E.-L. (2011). Syntactic and actional characteristics of Finn-
ish että-clauses. In R. Laury, & R. Suzuki (Eds.), Subordination in conversation: A cross-
linguistic perspective (pp. 69–102). Amsterdam: Benjamins.
https://doi.org/10.1075/slsi.24.05koi
Laury, R., & Helasvuo, M.-L. (2015). Detached NPs with relative clauses in Finnish conversa-
tion. In M. M. J. Fernandez-Vest & R. D. Van Valin (Eds.), Information structuring of spoken
language from a crosslinguistic perspective (pp. 149–166). Berlin: De Gruyter.
https://doi.org/10.1515/9783110368758-008
Laury, R., & Okamoto, S. (2011). Teyuuka and I mean as pragmatic parentheticals in Japanese
and EnglishIn R. Laury & R. Suzuki, Ryoko (Eds.), Subordination in Conversation: A Cross-
linguistic Perspective (pp. 209 238). Amsterdam, John Benjamins.
https://doi.org/10.1075/slsi.24.10lau
Laury, R., & Ono T. (2014). The limits of grammar: Clause combining in Finnish and Japanese
conversation. Pragmatics 24(3), 561–592. https://doi.org/10.1075/prag.24.3.06lau
Laury, R., & Suzuki, R. (Eds.). (2011). Subordination in Conversation: A Crosslinguistic Perspec-
tive. Amsterdam: Benjamins. https://doi.org/10.1075/slsi.24
Lindström, J., Lindholm, C., & Laury, R. (2016a). The interactional emergence of conditional
clauses as directives: Constructions, trajectories and sequences of actions. Language Sci-
Lindström, J., & Londen, A.-M. (2008). Constructing reasoning: The connectives för att (causal),
så att (consecutive) and men att (adversative) in Swedish conversations. In J. Leino (Ed.),
Constructional reorganization (pp. 105–152). Amsterdam: Benjamins.
https://doi.org/10.1075/cal.5.06lin
Lindström, J., Maschler, Y., & Pekarek Doehler, S. (Eds.). (2016b). Grammar and negative epis-
temics in talk-in-interaction: Cross-linguistic studies. Journal of Pragmatics 106, special issue.
Linell, P. (2009). Rethinking language, mind, and world dialogically: Interactional and contextual
theories of human sense-making. Charlotte, NC: Information Age Publishing.
Maschler, Y. (2009). Metalanguage in interaction: Hebrew discourse markers. Amsterdam: Ben-
jamins. https://doi.org/10.1075/pbns.181
Maschler, Y. (2012). Emergent projecting constructions: The case of Hebrew yada (‘know’).
Studies in Language 36(4), 785–847. https://doi.org/10.1075/sl.36.4.03mas
Maschler, Y. (2015). Word order in time: Emergent Hebrew (Ns)V/VNs syntax. In A. Dep-
permann, & S. Günthner (Eds.), Temporality in interaction (pp. 201–236). Amsterdam:
Benjamins.
Maschler, Y. (2017). The emergence of Hebrew loydea / loydat (‘I dunno masc/fem’) from inter-
action: Blurring the boundaries between discourse marker, pragmatic marker, and modal
particle. In A. Sansò, & C. Fedriani (Eds.), Pragmatic markers, discourse markers and modal
particles: New perspectives (pp. 37–69). Amsterdam: Benjamins.
https://doi.org/10.1075/slcs.186.02mas
Maschler, Y. (2018). The on-line emergence of Hebrew insubordinate she- (‘that/which/who’)
clauses: A usage-based perspective on so-called ‘subordination’. Studies in Language 42(3),
669–707.
Maschler, Y., & Dori-Hacohen, G. (2018). Constructing a genre: Hebrew lo yode'a / lo yoda'at
‘(I) don’t know’ on Israeli political radio phone-ins. Text & Talk 38(5), 575–604.
Maschler, Y., & Fishman, S. (forthcoming). From multi-clausality to discourse markerhood: The
Hebrew ma she- (‘what that’) construction in pseudo-cleft-like structures..
Matthiessen, C., & Thompson, S. A. (1988). The structure of discourse and ‘subordination’.
In J. Haiman, & S. A. Thompson (Eds.), Clause combining in grammar and discourse
(pp. 275–329). Amsterdam: Benjamins. https://doi.org/10.1075/tsl.18.12mat
Mondada, L. (2006). Participants’ online analysis and multimodal practices: Projecting the end
of the turn and the closing of the sequence. Discourse Studies 8, 117–129.
https://doi.org/10.1177/1461445606059561
Mondada, L. (2007). Multimodal resources for turn-taking: pointing and the emergence of pos-
sible next speakers. Discourse Studies 9(2), 194–225.
https://doi.org/10.1177/1461445607075346
Mondada, L. (2009). Emergent focused interactions in public places: A systematic analysis of the
multimodal achievement of a common interactional space. Journal of Pragmatics 41(10),
1977–1997. https://doi.org/10.1016/j.pragma.2008.09.019
Mondada, L. (2014a). Bodies in action: Multimodal analysis of walking and talking. Language
and Dialogue 4(3), 357–403. https://doi.org/10.1075/ld.4.3.02mon
Mondada, L. (2014b). The local constitution of multimodal resources for social interaction.
Journal of Pragmatics 65, 137–156. https://doi.org/10.1016/j.pragma.2014.04.004
Mondada, L. (2016). Challenges of multimodality: Language and the body in social interaction.
Journal of Sociolinguistics 20(3), 336–366. https://doi.org/10.1111/josl.1_12177
Norén, N., & Linell, P. (Eds.). (2013). Pivot constructions in talk in interaction. Journal of Prag-
matics 54, special issue.
Ochs, E., Schegloff, E. A., & Thompson, S. A. (Eds.). (1996). Interaction and grammar. Cam-
bridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511620874
Oloff, F., & Havlik, M. (2018). An initial description of syntactic extensions in spoken Czech.
Pragmatics 28(3), 361–390.
Pekarek Doehler, S. (2011a). Clause-combining and the sequencing of actions: Projector

constructions in French conversation. In R. Laury, & R. Suzuki (Eds.), Subordination in
conversation: A crosslinguistic perspective (pp. 103–148). Amsterdam: Benjamins.
Pekarek Doehler, S. (2011b). Emergent grammar for all practical purposes: The on-line format-
ing of dislocated constructions in French conversation. In P. Auer, & S. Pfänder (Eds.),
Constructions: Emerging and emergent (pp. 46–88). Berlin: de Gruyter.
https://doi.org/10.1515/9783110229080.45
Pekarek Doehler, S. (2015). Grammar, projection and turn-organization: (il) y a NP ‘there is
NP’ as a projector construction in French talk-in-interaction. In A. Deppermann, & S.
Günthner (Eds.), Temporality in interaction (pp. 173–200). Amsterdam: Benjamins.
Pekarek Doehler, S. (2016). More than epistemic hedge: Je sais pas ‘I don’t know’ in French
talk-in-interaction. Journal of Pragmatics 106, 148–162.
Pekarek Doehler, S. (2019). At the interface of grammar and the body: chai pas (‘dunno’) as
a resource for dealing with lack of recipient response. Research on Language and Social
Interaction. 52(4). https://doi.org/10.1080/08351813.2019.1657276.
Pekarek Doehler, S., De Stefani, E., & Horlacher A.-S. (2015). Time and emergence in grammar:
Left-dislocation, right-dislocation, topicalization and hanging topic in French. Amsterdam:
Benjamins. https://doi.org/10.1075/slsi.28
Pekarek Doehler, S., & Horlacher, A.-S. (2013). The patching together of pivot-patterns in talk-
in-interaction: On ‘double dislocations’ in French. Journal of Pragmatics 53, 92–108.
Polak-Yitzhaki, H., & Maschler, Y. (2016). Disclaiming understanding? Hebrew 'ani lo mevin/a
(‘I don’t understand masc/fem’) in everyday conversation. Journal of Pragmatics 106,
Ross, J. R. (1969). Guess who? In R. I. Binnick, A. Davidson, G. M. Green, & J. L. Morgan (Eds),
Papers from the fifth regional meeting of the Chicago Linguistic Society (pp. 252–286). Chi-
cago, IL: The Chicago Linguistic Society.
Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization of
turn-taking for conversation. Language 50, 696–735. https://doi.org/10.1353/lan.1974.0010
Schegloff, E. A. (1979). The relevance of repair for syntax-for-conversation. In T. Givón (Ed.),
Syntax and semantics 12: Discourse and syntax (pp. 261–288). New York: Academic Press.
Schegloff, E. A. (1996). Turn organization: One intersection of grammar and interaction. In E.
Ochs, Schegloff, E. A., & Thompson, S. A. (Eds.), Interaction and grammar. Cambridge:
Cambridge University Press. https://doi.org/10.1017/CBO9780511620874.002
Selting, M., & Couper-Kuhlen, E. (Eds.) 2001. Studies in interactional linguistics. Amsterdam:
Benjamins. https://doi.org/10.1075/sidag.10
Stoenica, I.-M. (2014). Répétition et différenciation dans les reprises structurelles intégrant des
relatives. TRANEL 60, 209–220.
Stoenica, I. M. (2018). Le rôle des relatives dans l’organisation séquentielle de l’interaction: une
approche temporelle, interactionnelle et praxéologique. (Doctoral thesis.) Neuchâtel: Univer-
sity of Neuchâtel.
Sorjonen, M. L., & Raevaara, L. (2014). On the grammatical form of requests at the convenience
store: Requesting as embodied action. In E. Couper-Kuhlen, & P. Drew (Eds.), Requesting
in social interaction (pp. 243–268). Amsterdam: Benjamins.
Streeck, J. (2002). Grammars, words, and embodied meanings: On the uses and evolution of so
and like. Journal of Communication 52(3), 581–596.
https://doi.org/10.1111/j.1460-2466.2002.tb02563.x
Streeck, J., Goodwin, C., & LeBaron, C. (Eds.). (2011). Embodied interaction: Language and
body in the material world. Cambridge: Cambridge University Press.
Streeck, J., & Jordan, J. S. (2009). Projection and anticipation: the forward-looking nature of
embodied communication. Discourse Processes, 46(2–3), 93–102.
Tomasello, M. (2003). Constructing a language: A usage-based theory of language acquisition.
Cambridge, MA: Harvard University Press.
Thompson, S. A. (2002). “Object complements” and conversation: Towards a realistic account.
Studies in Language 26, 125–163. https://doi.org/10.1075/sl.26.1.05tho
Thompson, S. A., & Couper-Kuhlen, E. (2005). The clause as a locus of grammar and interac-
tion. Discourse Studies 7(4–5), 481–505. https://doi.org/10.1177/1461445605054403
Thompson, S. A., & Mulac, A. (1991). A quantitative perspective on the grammaticization of
epistemic parentheticals in English. In E. Closs Traugott, & B. Heine (Eds.), Approaches to
grammaticalization (pp. 313–329). Vol. II. Amsterdam: Benjamins.
https://doi.org/10.1075/tsl.19.2.16tho
Van Valin, R. D. (1984). A typology of syntactic relations in clause linkage. Proceedings of the
Tenth annual meeting of the Berkeley Linguistic Society, 542–558.
Walker, G. (2007). On the design and use of pivots in everyday English conversation. Journal of
Pragmatics 39, 2217–2243. https://doi.org/10.1016/j.pragma.2006.10.002
Wide, C. (2014). Constructions as resources in interaction: Syntactically unintegrated att ‘that’-
clauses in spoken Swedish. In R. Boogaart, T. Colleman, & G. Rutten (Eds.), Expanding the
scope of construction grammar (pp. 353–388). Berlin: Mouton de Gruyter.
Part I
Emerging projecting constructions

Chapter 2
Nel senso (che) in Italian conversation

Turn-taking, turn-maintaining and turn-yielding
Elwys De Stefani
This chapter examines nel senso (che) ‘in the sense (that)’ in present-day
conversational Italian. Speakers can use the resource to start a turn, to extend a
turn-in-progress, or to yield a turn to a next speaker. In TCU-beginnings, it is
describable as a projector construction, allowing speakers to display continuation
of their turn (with a new clause). When extending a turn with nel senso (che),
speakers display orientation towards a potential problem of understanding
which they prevent by elaborating on their prior talk. In turn- and TCU-initial
positions, both nel senso che and nel senso are observed; in turn-final positions
only the latter format occurs. The chapter analyzes the syntactic, prosodic,
embodied, and praxeological corollaries of the resource.
Keywords: turn position, projector construction, Italian, syntax
1. Introduction
This chapter examines the resource nel senso (che) – literally ‘in the sense (that)’ –
which has been described as “more and more used in spoken Italian” (Dal Negro
& Fiorentini, 2014, p. 102) and which sometimes causes irritation among speakers
who observe its pervasive use.1 Traditionally, nel senso che is observed in construc-
tions where the relative pronoun che introduces a relative clause which modifies
1. In two short notes published by the national newspaper Corriere della Sera on February 12
and 25, 2017 (accessible at https://tinyurl.com/ydcbbq6k and https://tinyurl.com/ya2wn8m2),
the Italian journalist Paolo Di Stefano documents the use of nel senso (che) among students and
young people. Earlier evidence dates back to June 6, 2005, when a forum participant criticizes
the widespread use of nel senso (che) (http://tgmonline.gamesvillage.it; no longer accessible);
incidentally, the user seems to live in the province of Milan, which many speakers see as the area
from which this use has spread (see also Fiorentini & Sansò, 2017, p. 69). [All links accessed
May 27, 2018].
https://doi.org/10.1075/slsi.32.02ste
26 Elwys De Stefani
the noun phrase (NP) senso. In Excerpt (1), the NP senso ‘sense’ constitutes an
antecedent ‘head noun’ for the subsequent relative clause:
(1) Giacomo Leopardi, Zibaldone, August 26, 1821
L’individuo non è virtuoso, la moltitudine sì, e sempre, per le ragioni e nel senso
che ho sviluppato altrove.
‘The individual is not virtuous, the multitude is, and always, for the reasons
and in the sense that I have developed elsewhere.’
nel senso che ho sviluppato altrove.
in.the sense rel have-prs.1sg develop-ptcp.prf elsewhere
‘in the sense that I have developed elsewhere’
antecedent relative clause
Whereas the abovementioned use is still observed in Italian, this chapter zooms in
on occurrences of nel senso che in which che is not used as a relativizer modifying an
antecedent but rather as part of a multi-word conjunction. The following excerpt is
taken from a Roman dinner conversation among four colleagues and friends who
are discussing the reasons why a joint acquaintance has recently changed her job:
(2) Ro13CE2–11 (09: 59–10: 06)2
01 VER: >secondo me< forse aveva qual[che::
in my view perhaps she had some
02 GIU: [ANche ha detto
also she said
03 per calcolo.
by calculation
04 nel senso che lì non avrebbe avuto
in the sense that there she wouldn’t have had
05 progressione di carr[iera e invece di qua c’ha un&
career progression whereas here she has a
06 SON: [è vero.
that’s true
07 GIU: &tizio che va in pensione,
guy who is going to retire
Here, Giulia (GIU) affirms that her acquaintance has changed her job ‘by calcula-
tion; out of self-interest’ (l. 03). She then extends her turn with the words nel senso
che, thereby introducing the grounds for such a ‘calculation.’ In this case, nel senso
2. Talk has been transcribed according to Jefferson’s (2004) conventions. The symbol ‘&’
is added to the end of a line of talk that is overlapped and indicates turn-continuation at the
next line beginning with the same symbol. The symbol ^ indicates liaison. The annotation of
embodied conduct follows Mondada’s (2016) guidelines.
Chapter 2. Nel senso (che) in Italian conversation 27
che is a multi-word conjunction that enables Giulia to extend a possibly complete

turn-constructional unit (TCU; Sacks, Schegloff, & Jefferson, 1974), identifiable as
such thanks to the falling intonation on “calcolo.” (l. 03) as well as to the syntactic
and pragmatic completeness of the turn-so-far.
nel senso che lì non avrebbe avuto progressione di carriera
in.the sense conj there not have-cond.3sg have−prcp.prf progression of career
‘in the sense that there she wouldn’t have had career progression’
conjunctive phrase clause
These illustrations show regular manifestations of che in Italian, which may be

used as a pronoun introducing a relative clause (Excerpt 1), or as a conjunction.3
As such, che can occur as a full-fledged conjunction4 or as part of a conjunctive
phrase (“locuzione congiuntiva”; see Serianni, 1988, p. 305), as in Excerpt (2). This
chapter zeroes in on the latter use.
2. Previous studies
In Italian, numerous conjunctive phrases include the element che.5 Whereas most
of these are recorded in dictionaries, nel senso che is largely ignored by contem-
porary lexicography. For instance, it is not mentioned in the recent edition of
one of the leading dictionaries of Italian language (Zingarelli, 2018).6 Although
frequently occurring in spoken Italian, nel senso che has received only scarce
attention so far. The same holds true for the (truncated) form nel senso, which
Fiorentini and Sansò (2017) describe as a reformulation marker. Acknowledging,
however, that “it may have other uses in which reformulation is not involved”
(Fiorentini & Sansò, 2017, p. 62), the authors note that nel senso can be used in
turn-initial, turn-medial, or turn-final positions. At turn-beginning, nel senso is
said to introduce a “new discourse unit”; mid-turn uses of nel senso are described
as initiating “reformulation” of the utterance so far, or as “emphasizing and
3. Che can also be used as an interrogative adjective (according to the Italian terminology),
translatable as ‘what’, or ‘which.’
4. A typical conjunctive use of che would be prendi l’ombrello che piove! ‘take the umbrella
[since, because, given that] it’s raining!’ (Dardano & Trifone, 1997, p. 378).
5. See for instance dato che ‘given that,’ di modo che ‘in a manner that,’ ‘so that,’ al punto che ‘to
the point that,’ ‘to the extent that,’ as well as the graphically contracted forms sicché (sì che) ‘so
that,’ benché (bene che) ‘although,’ among many others.
6. Incidentally, but relevantly, it is worth noting that the authors of the dictionary use nel senso
che 11 times in their explanations of the lemmata.
28 Elwys De Stefani
monitoring” a preceding portion of talk; turn-final uses appear to achieve “boost-

ing” or “hedging” of previous talk or to do “insinuating,” whereby “the speaker
invites the hearer to draw all the common-knowledge or context-driven inferences
connected to the previous statement” (Fiorentini & Sansò, 2017, p. 65). Whereas
the authors show that nel senso is used recurrently to connect portions of talk, the
syntactic and turn-constructional facets of such linkages are only marginally ad-
dressed. In fact, the authors’ positional and functional description of nel senso does
not enable us to understand how speakers use it in the organization of interaction.
In this chapter, I propose an analysis of the grammatical features of nel senso (che)
that focuses also on the speakers’ embodied conduct, thereby revealing how this
conjunctive format is used as a resource to manage turns-at-talk.
3. Data and method
The analysis is based on a collection of 19 instances of nel senso and 25 instances of

nel senso che, which I have collected from various settings of both ordinary dinner-
table conversations and institutional interaction (work meetings, driving lessons,
classroom interaction, travel agency consultations). The data were collected
between 2010 and 2013 in various parts of the Italian-speaking area, ranging from
Ticino, Switzerland to Naples in southern Italy.7 Indebted to conversation analysis
(Sacks et al., 1974) and interactional linguistics (Selting & Couper-Kuhlen, 2001;
Couper-Kuhlen & Selting, 2018), I first look at how speakers use nel senso (che)
at the beginning of responsive turns (Section 4). In the subsequent section, I con-
sider its use in turns-in-progress (Section 5), whereas the final section zooms in on
turn-final uses of nel senso (Section 6). The study focuses on how (grammatical)
turn-construction relates to the organization of actions. Interactional research on
grammar has shown that grammar enables speakers to implement and organize
socially relevant actions (Ochs, Schegloff, & Thompson, 1996; Thompson &
Couper-Kuhlen, 2005). At the same time, several authors have problematized the
ways in which grammatical units (e.g., clauses) relate to interactional units, such
7. This article is a contribution to the project Beyond the clause: Encoding and inference in clause
combining funded by KU Leuven (project number C14/18/034) and co-directed by Bert Cornillie,
Kristin Davidse, Elwys De Stefani and Jean-Christophe Verstraete (2018–2022). The data used
for this study were collected within two projects: 1) The constitution of space in interaction: A
conversation analytic approach to the study of place names and spatial descriptions, financed by
the Swiss National Science Foundation (project number PP001–119138) and directed by Elwys
De Stefani at the University of Berne (2008–2012); 2) ALIAS (Archivio di LInguA Spontanea),
funded by KU Leuven as an educational project (project number OWP2012/08) and directed by
Stefania Marzo, Elwys De Stefani and Bart Van Den Bossche (2012–2014).
as TCUs (e.g., Couper-Kuhlen, 2012; Lindström, 2006; Selting, 2005), especially

with respect to increments or other turn-extensions occurring after a transition
relevance place (TRP; Sacks et al., 1974). These issues are at the core of research on
clause-combining from an interactional perspective, and are central to the present
discussion of nel senso (che).
4. Turn-initial nel senso (che) in responsive actions
Turn-initial nel senso (che) occurs in second pair parts produced after questions
such as in che senso? (‘in what sense?’), as the following two fragments show.
Excerpt (3) is taken from a driving lesson: at lines 01–02, the instructor (INS)
articulates a general rule that the trainee driver (TDV) is expected to follow. The
latter receives this utterance with cioè in che senso. ‘that is in what sense’ (l. 04),
thereby initiating a repair sequence. The instructor does not immediately respond
to the question but rather produces another instruction related to the traffic situ-
ation at hand, which the trainee driver is expected to respond to immediately (ll.
06–07). After having assessed the execution of that instruction (l. 15), INS articu-
lates the second pair part of the repair sequence (ll. 17–21):
(3) 15sg2BM3 (41: 06–41: 33)
01 INS: non girare mai il volante, (.) s:e non: non capisci
never turn the steering wheel (.) if you don’t make
02 la direzione nella quale tu stai andando eh?
out the direction in which you are going right
03 (1.1)
04 TDV: [cioè in che senso.
that is in what sense
05 INS: [ə-
06 INS: adesso: facciamo la svolta qui per scendere a
now we turn here to go down the
07 sinistra poi te:-
left-hand side and then
((8 lines omitted))
15 INS: bravissimo.
well done
16 (1.0)
17 INS: nel senso se tu sei fermo a uno stop, (0.7) non
in the sense if you stand at a stop (0.7) don’t
18 girare il volante così. (.) per già
turn the steering wheel like that (.) to already
30 Elwys De Stefani
19 mettere le ruote s[enza che tu mettaîn&

set the wheels without bringing
20 TDV: [ah. (.) (°° °°)
oh
21 &leggero movimento la macchina.
the car into slight movement
22 TDV: mhm.
The turn with which the instructor completes the repair sequence starts with
the words nel senso ‘in the sense.’ By recycling the word senso ‘sense’ – which the
trainee driver had used in his repair initiation (l. 04) – the instructor is displaying
with the very first words of his turn that he is now completing the sequence. In this
case, starting the TCU with nel senso allows the instructor to display that he is now
resuming an action he had suspended temporarily for the benefit of a different
action (here: providing an instruction); in other words, he displays a link back (De
Stefani & Horlacher, 2008) to the start of a sequence-to-be-completed.
The next excerpt is taken from a service encounter between a travel agent
(Carolina, CAR) and a couple, Pedro (PED) and Pina (PIN), who wish to suggest a
trip to a couple of friends and who ask Carolina about how to proceed, given that
they cannot indicate fixed days on which the trip should take place:
(4) Na09AV3A-1 (48: 12–48: 32)
01 PED: sì ə: carolì ma non si può fare una:: data libera.
yes ehm Carolì but can’t we make a free date
02 (.)
03 PED: si può fare solamente il buono?
can one just make the voucher
04 (1.6)
05 CAR: n:o. (.) nel senso che noi possiamo pure fare una
no (.) in the sense that we can also do a
06 cosa tra di noi: (.) e basta. (.) e
thing between us (.) and that’s it (.) and
07 no[i mettiamo wee ]kendə di giugno.&
we put the weekend of June
08 PED: [°in che senso.°]
in what sense
09 CAR: &nel senso che. (1.2) sappiamo che suppergiù il
in the sense that (1.2) we know that more or less
10 prezzo è quello.
that’s the price
11 (.)
12 PED: hm.
13 CAR: senza fare nulla e allora gli diciamo il periodo di
without doing anything and then we tell them the
14 giugno weekend venerdì domenica. gli mettiamo le

period of June weekend Friday Sunday we allocate
15 due notti.
them the two nights
16 (0.3)
17 CAR: e loro decidono.
and they decide
At lines 01–03 Pedro asks Carolina whether it would be possible to fare ‘make’
a voucher without indicating the precise dates of the trip. Note that Carolina
responds first with a no (l. 05) before elaborating on her response by initiating
the subsequent TCU with nel senso che (see Section 5). However, Pedro exhibits
understanding problems with respect to Carolina’s words, as he initiates repair at
line 08 by overlapping her talk with a softly spoken °in che senso.° ‘in what sense.’
Carolina ends her overlapped turn (as the falling intonation on giugno. ‘June’
shows; l. 07) and then immediately initiates the completion of the repair sequence
by saying nel senso che. (l. 09). Several aspects of the use of nel senso che. at line
09 are noteworthy: Prosodically, the element che. is emphasized and produced
with a falling intonation; this prosodic configuration is highly projective, as the
subsequent 1.2-second pause shows, which is not heard as an occasion for turn
transition. In other words, this occurrence of nel senso che. can be described as
a projector phrase (Günthner, 2011) or a projector construction (Pekarek Doehler,
2011), in that it cannot constitute on its own the expected second pair part but
rather projects that the answer to Pedro’s repair initiation is coming up next.
Finally, this occurrence shows a case in which the speaker produces nel senso che.
as a contingently emerging continuation of her turn-in-progress, which she adds
without any gaps or hitches to the end of her previous TCU (giugno. ‘June’; l. 07).
Syntactically speaking, Carolina can be seen to combine two clauses in a smooth
way by using the resource nel senso che. – but this clause combination emerges
from a micro-sequential adjustment of her turn-in-progress to Pedro’s overlapping
talk and action. In other words, Carolina first completes her answer (ll. 05–07) to
Pedros’s preceding question (l. 03) and then engages in a different action with the
words nel senso che. (l. 09), namely responding to Pedro’s repair initiation (l. 08).
Nel senso can also be used as a turn-entry device (Sacks et al., 1974, p. 719) in
other kinds of responsive actions. The following excerpt shows a case in which a
speaker engages in speakership, but does not complete her turn. It is taken from a
classroom interaction collected in Naples during a history lesson. The schoolgirl
(SCG) has just read a piece of text aloud and the teacher (TEA) repeats the expres-
sion proiettarci nel futuro ‘to project ourselves into the future’ (l. 01) that was part
of the schoolgirl’s reading and now asks what it means:
32 Elwys De Stefani
(5) Ca13CL2–10 (20: 12–20: 28)

01 TEA: proiettarci nel futuro che significa.
to project ourselves into the future what
does that mean
02 (0.9)*(0.2)*(0.2)*
TEA *.....*gazes at SCG*-->
03 SCG: ə::::
er
04 (0.5)
05 TEA: *cosa* vuol dire proiettarsi nel futuro.
what does it mean to project oneself into the future
TEA -->*,,,,*
06 (1.4)*(0.2)*
TEA *.....*-->
07 SCG: *nel senso:::*
in the sense
TEA -->*gazes at SCG*-->>
08 (2.6)
09 TEA: assicurare un futuro all’umanità.
to ensure a future for humanity
10 (0.5)
11 SCG: eh sì.
well yes
The teacher’s question (l. 01) is potentially addressed to all the pupils present in
the room, but her gaze towards the schoolgirl who has just read the text (l. 02)
is treated by the latter as selecting her as the next speaker (Stivers & Rossano,
2010). However, she produces only a lengthened ə:::: (l. 03), after which the teacher
withdraws her gaze while reproducing her question (l. 05). It is upon the teacher’s
second gaze reorientation towards the girl that the latter utters a lengthened nel
senso::: ‘in the sense’ (l. 07), thereby displaying her understanding of herself as
the next speaker. She fails however to provide an answer, and the teacher herself
eventually completes the sequence shortly thereafter (l. 09). This excerpt shows
the use of nel senso::: as displaying the schoolgirl’s right and obligation to provide
an answer to a display question (Lee, 2006). The 2.6-second pause (l. 08) shows
the potential projective strength of the resource, which the teacher will eventually
cut short by providing the answer to her own question. Contrary to what we have
seen in Excerpts (3) and (4), in this case the teacher’s question-word question
problematizes the meaning of the expression proiettarsi nel futuro ‘to project
oneself into the future’ (l. 05). The fact that the teacher asks a question about the
‘meaning’ of a specific expression provides a rationale for why the schoolgirl uses
nel senso::: ‘in the sense,’ which possibly orients precisely to the meaning-problem
the teacher makes relevant. An orientation towards a potential problem of mean-

ing and understanding is indeed at stake also in other uses in which nel senso (che)
appears at TCU-beginnings, as the following analyses will show.
5. Nel senso (che) and turn-construction
On occasion, speakers may extend a turn beyond a place of possible completion by

using nel senso (che). For extensions that are grammatically fitted to prior talk and
heard as recompleting the prior turn, Schegloff (1996) introduced the notion of
increment.8 The conjunctive phrase nel senso (che) appears to exhibit grammatical
dependence on prior talk, while at the same time projecting more talk to come.
5.1 Turn-extensions after other-speaker’s talk
Schegloff (2016, p. 253) observed that increments may occur “post-other-talk.”

The following excerpt provides a case in point. The travel agent Carolina (see
Excerpt 4) is giving advice to Gino (GIN), a client who wishes to visit diamond fac-
tories. They have just discussed the possibilities of buying diamonds in Mauritius.
At lines 01–05, Carolina introduces information that is brand new for Gino: She
explains that whatever he buys at the different sites in Mauritius will be kept for
him and delivered to him only at the airport on his way back:
(6) NA09AV3–1 (05: 49–06: 10)
01 CAR: e poi la cosa fondamentale tutto ciò che viene
and then the fundamental thing everything that is
02 acquistato lì:,.h non ve lo fanno portare.
bought there .h they don’t let you take it along
03 GIN: hm.
04 CAR: lo ricevete direttamente in aeroporto.
you receive it directly at the airport
05 al vostro ritorno.
on your way back
06 (1.4)
07 GIN: ah.
oh
08 *(1.0)*(0.3)
CAR *nods-*
8. For an overview of the terminological and conceptual variety with which turn-extensions are
described, see Couper-Kuhlen & Selting (2018, pp. 94–103).
34 Elwys De Stefani
09 CAR: nel senso che. (.) voi vedete quello che acquistate
in the sense that (.) you see what you buy
10 fate tutto.
you do everything
11 (0.7)
12 CAR: non vi danno nulla.
they don’t give you anything
13 GIN: ‘eh.(.) ma uno paga?
right (.) but one pays
14 (0.3)
15 CAR: pagate.
you pay
16 (0.2)
17 GIN: ‘eh. (.) va [beh.
right (.) well okay
18 CAR: [fate tutto. non vi danno nulla
you do everything they don’t give you
19 ricevete tutto in aeroporto.
anything you receive everything at the airport
Carolina delivers the new piece of information in several steps. Her first TCU (ll.
01–02) is followed by Gino’s short acknowledgement (l. 03), after which Carolina
extends her turn (l. 04). After al vostro ritorno. ‘on your way back,’ a 1.4-second gap
occurs, which shows that turn-transition is expectable here (l. 06). However, Gino
produces only the change-of-state token ah. ‘oh’ (Heritage, 1984). Carolina can be
seen as treating Gino’s response as insufficient: a further gap occurs (l. 08), during
which Carolina nods. At line 09, Carolina self-selects with the words nel senso che.
‘in the sense that,’ followed by a brief pause, after which she produces a second
version of the information she has just provided (ll. 09–19). By doing so, Carolina
shows that she treats Gino’s delayed receipt token ah. ‘oh’ as possibly displaying an
understanding problem. Whereas the speakers in Excerpts (3) and (4) displayed
their understanding problem overtly by saying in che senso? ‘in what sense?’, here
Carolina treats the minimal and delayed way in which Gino responds to her in-
forming as hinting at a potential understanding problem on his part. Continuing
with nel senso che. allows her to display that she is going to tackle that problem.
As in Excerpt (4), nel senso che. in (6) is articulated with a falling intonation,
with stress on che., and followed by a pause. These prosodic and temporal features
seem to challenge the common conversation analytic understanding according
to which an intonation contour which falls to low is heard as “utterance-final”
(Jefferson, 2004). But nel senso che. is clearly syntactically incomplete, and the fol-
lowing pause is heard as a “holding silence” (Local & Kelly, 1986), after which the
speaker will continue her turn.
5.2 Turns-in-progress
In this section, I focus on instances in which a speaker reaches a TRP and then
extends his or her turn with nel senso (che), thereby building a multi-unit turn.
Whether the projected extension is heard as elaborating on a prior action or as
starting a new action is the central question discussed in this section.
5.2.1 Elaborating on a previous action

The following excerpt is taken from a business meeting that took place in Milan.
Annina (ANN) is presenting her financial report and the chairman Paolo (PAO)
has just asked for some clarification about offers that their company had requested.
Annina starts her reply in the following way:
(7) Mi13PR1–42 (24: 36–24: 51)
01 ANN: posso dirti: a stima che::^hm: (1.5)
I can tell you roughly that hm (1.5)
02 dalle sollecitazioni offerte^è
from the offer requests
03 venuto fuori: pocoô nulla.
nothing or little has come out
04 nel senso::^hm è venuta soltanto fuori
in the sense hm the only thing that came out
05 non si vede nelle offerte ma nella
one doesn’t see it in the offers but in the
06 pianificazione, una questione legata ad
planning is a question related to
07 ((company name)).
((company name))
Annina starts her answer with what becomes recognizable as a clausal preface:
whereas her first TCU (posso dirti: … pocoô nulla ‘I can tell you …has come out’;
ll. 01–03) is syntactically and prosodically complete, as the downward intonation
on nulla. ‘nothing’ shows, it appears to be a pragmatically insufficient answer
to the chairman’s question. In other words, by extending her own turn, Annina
exhibits that what she has said so far is not to be heard as the complete answer
to Paolo’s question. She extends her turn-in-progress immediately by saying nel
senso::^hm (l. 04). This extension is prosodically disjunctive with respect to the
falling intonation of the previous unit (nulla. ‘nothing’; l. 03).9 Syntactically, nel
senso::^hm combines the following two clauses:
9. I call “prosodically disjunctive” those combinations in which there is “no common line of
pitch declination” (Couper-Kuhlen, 2012, pp. 279).
36 Elwys De Stefani
clause 1: dalle sollecitazioni offerte^è venuto fuori:

from requests offers be-prs.3sg come-ptcp.prf out
pocoô nulla.
little or nothing
nothing or little has come out from the offer requests
nel senso::^hm
in.the sense
in the sense
clause 2: è venuta soltanto fuori [parenthetical] una
be-prs.3 ps come-ptcp.prf only out [parenthetical] a
questione legata ad ((company name))
question tie-ptcp.prf to ((company name))
the only thing that came out is a question related to ((company name))
Excerpt (8) shows that not only nel senso but also nel senso che can be used as
a resource for clause- and TCU-combining. The excerpt is taken from the same
business meeting and occurs some 20 minutes later. The team is discussing the
reorganization of tasks among the colleagues. Paolo, the chairman, has previ-
ously asked how the business travels will be reorganized. We join the discussion as
Pamela (PAM) explains her plan:
(8) Mi13PR1–42 (45: 58–46: 13)
01 PAM: e poi ci *sarà:*: (.) cioè ci stiamo organizzando
and then there’ll be (.) that is we’re organizing
PAO *.....*gazes at PAM-->
02 perché ci sarà un forte lavoro iniziale.#
ourselves because there’ll be tough initial work
fig #fig. 1
03 nel senso che bisogna::* (0.4)*
in the sense that we need to (0.4)
PAO -->*,,,,,,*
04 +pensare *bene* alle conven*+zio#ni, (0.5)
think out the conventions (0.5)
PAO *....*gaze at PAM-*
PAO +slight nodding-------------+vigorous nodding-->
fig #fig.2
05 #pens*are *bene+ *a:*+:: ai vari s*posta*menti*
think out the the various movements
PAO *....*gaze PAM*,,*gaze PAM-----*,,,,,*PAM--*
PAO -->+,,,,,,,+
fig #fig.3
06 *quindi* ci sarà +proprio
hence there will be really
PAO *......*gaze withdrawn from PAM-->
PAO +slight nodding-->>
07 un for*te* .h (.) pensiero (.) a(l) priori.*

a strong thought a priori
PAO -->*..*gaze at PAM----------------------*
We observe again similar turn-constructional features as in Excerpt (7): Pamela

produces a syntactically complete clause ending with a falling intonation (iniziale.
‘initial’; l. 02) and then extends her turn, this time using the form nel senso che:
clause 1: ci sarà un forte lavoro iniziale.
there be-fut.3sg a strong work initial
there will be tough initial work
nel senso che
in.the sense conj
in the sense that
clause 2: bisogna:: (0.4) pensare bene alle convenzioni, ((continues)).
need-prs.3sg think well to.the conventions ((continues))
we need to think out the conventions ((continues))
In this case too, the speaker treats the clause preceding the turn-extension as
pragmatically incomplete, the added clause hence providing additional informa-
tion – in this case exemplification – of what she means by ‘tough initial work.’ But
why is the speaker treating her own talk as pragmatically insufficient? On the one
hand, at the end of line 02, Pamela has announced what participants can under-
stand as ‘a lot of work’ without mentioning of what that work will consist.10 On the
other hand, the embodied conduct of Pamela’s primary addressee (the chairman
Paolo) shows that there are interactional reasons for extending a turn-in-progress
in this way at this precise moment. Indeed, Paolo is looking at Pamela in a rather
Figure 1. Paolo (and other participants) looking at Pamela
10. Note too that both in Excerpt (7) (‘little or nothing’) and in Excerpt (8) (‘tough initial work’)
the clause preceding the nel senso (che)-elaboration contains the expression of an approximate
quantification.
38 Elwys De Stefani
static way during the initial parts of her turn (Figure 1), without displaying any
embodied response to what Pamela is saying. I argue that Pamela’s turn-extension
is sensitive to the absence of embodied response by her primary recipient. Indeed,
by extending her turn, Pamela manages to receive an embodied response from
Paolo, who first slightly, then vigorously, nods while listening to Pamela’s words
(Figures 2 and 3).
Figure 2. Paolo vigorously nodding Figure 3. Paolo vigorously nodding
Excerpts (7) and (8) can be described as “next-beat increments” (Schegloff, 2016),
whereby speakers exhibit the pragmatic incompleteness of their turn-so-far with
increments. By extending their turn with nel senso (che), speakers seem to orient
towards possible imprecise or insufficient understanding that the turn-so-far might
entail. From a syntactic perspective, speakers can be seen to combine two clauses.
On occasion, however, what follows nel senso (che) is not a new clause, as in
the following excerpt taken from the same meeting. Paolo is addressing all his
teammates as he asks them to think about possible financial cuts:
(9) Mi13PR1–43 (18: 21–18: 42)
01 PAO: io vi chiedo:, (0.7)
I am asking you(0.7)
02 ogni: (.) non ogni due per tre. (.) però:: (1.3)
every (.) not every once in a while (.) but (1.3)
03 di p:ensare, (.) ad altri possibili tagli. (1.5)
to think (.) of other possible cuts (1.5)
04 nel duringə.11 (0.5)
in the course of the process (0.5)
05 nel senso che (0.6)
in the sense that (0.6)
06 tagli (0.4) come abbiamo scritto ieri sera nella
cuts (0.4) as we’ve written yesterday evening in
07 comun- nel comunicato, .hh che non inficino una
the comm- in the communiqué that don’t undermine a
11. Paolo uses the English preposition during as a substantive: nel duringə ‘in the during’ can be
understood as ‘in the course of the process.’
08 tras- stra- strategia di sviluppo.

tras- stra- strategy of development
Numerous long pauses occur during Paolo’s talk. These are not heard as TRPs,
since none of the four co-present participants (see Figure 1) attempts to gain
speakership. Paolo is addressing a request to his collaborators that is syntactically
and prosodically complete (‘I am asking you to think of other possible cuts in the
course of the process’; ll. 01–04). The end of line 04 is not treated as an occasion
for turn-transition, or as an opportunity to display embodied responses. Paolo
extends his turn with nel senso che (l. 05), after which he again allows a pause to
occur. He continues his turn by recycling the previously used (l. 03) NP tagli ‘cuts,’
which he now specifies as cuts che non inficino una tras- stra- strategia di sviluppo
‘that don’t undermine a development strategy’ (ll. 06–08). What follows nel senso
che here is thus an NP (tagli ‘cuts’; l. 06), a parenthetical (‘as we’ve written yesterday
in the communiqué,’ ll. 06–07), and a relative clause extension (ll. 07–08).
This excerpt shows that the projector construction nel senso (che) can be fol-
lowed by language material that is not describable as an independent main clause.
However, the fact that nel senso che is used here to specify the intended meaning of
tagli ‘cuts’ shows, once again, that the speaker orients towards a possible imprecise
understanding of his turn-so-far.
In all the cases analyzed in this section, nel senso (che) allows speakers to
introduce an explanation or specification of a possible problem of understanding
related to their own previous talk. What follows nel senso (che) in excerpts 6 to
9 is not a new action but rather an elaboration on a still-relevant action, which
is done in the service of securing an appropriate understanding of what the
speaker is saying (see Deppermann and De Stefani, 2019; Stoenica and Pekarek
Doehler, this volume).
These elaborations are not syntactically or prosodically projected by the pre-
ceding talk, which qualifies them as turn expansions (Auer, 2009). The following
section describes an occurrence in which the resource introduces not only a new
clause but also a recognizably different action.
5.2.2 Initiating a new action

The fragment is taken from the same business meeting: Annina is presenting her
report and is mentioning the company’s ‘marketing structure,’ announcing that
this issue will be discussed on the following day:
(10) Mi13PR1–43 (13: 23–13: 37)
01 ANN: un altro discorso riguardava la struttura
another matter had to do with the marketing
40 Elwys De Stefani
02 marketing. (0.4) di questa parleremo domani. (.)

structure (0.4) we’ll talk about this tomorrow (.)
03 nel senso che non è stato fatto (.) per ora (.)
in the sense that there hasn’t been (.) for now (.)
04 paolo (.) nessun tipo di ə::: discorso
Paolo (.) any sort of er discussion
05 cred[o (eh)?
I believe right
06 PAO: [no: no. (.) viene fatto domani.
no no it’s going to be done tomorrow
The prosodic and syntactic turn-constructional features of Annina’s talk are simi-
lar to what we have observed in Excerpts (7) and (8):
clause 1: di questa parleremo domani.
of this-fem talk-fut.3pl tomorrow
we will talk about this tomorrow
nel senso che
in.the sense conj
in the sense that
clause 2: non è stato fatto (.) per ora (.) paolo (.)
not be-prs.3sg be-ptcp.prf make-ptcp.prf for now Paolo
nessun tipo di ə::: discorso
no.one type of discourse
there hasn’t been for now Paolo any kind of discussion
Annina reaches the end of a TCU as she says ‘we will talk about this tomorrow’
(l. 02). This TCU is complete not only with regard to prosody and syntax but also
pragmatically, namely with respect to the action that Annina accomplishes, i.e.,
providing information about the ‘marketing structure.’ Indeed, what follows nel
senso che is not hearable as an explanation of her previous talk (in other words,
Annina is not orienting towards a possible understanding problem related to
‘we will talk about this tomorrow’). Rather, she initiates a different action: She
articulates a declarative question and selects Paolo as the recipient who is able and
entitled to confirm her request. At line 06, Paolo does so in a type-conforming
(Raymond, 2003) manner. This way of extending a turn-in-progress is similar to
what Local and Walker (2004) have called abrupt-joins, a resource enabling speak-
ers to produce “multi-unit, multi-action turns” (p. 1377).
5.3 Self-repair
The recurrence of nel senso (che) in self-repair is probably what has led linguists
to treat the device as a reformulation marker (Fiorentini & Sansò, 2017). The
following excerpt provides a case in point. It is taken from the same interaction as
Excerpt (4): Pedro and Pina are asking the travel agent Carolina about a trip with
a couple of friends. Carolina is suggesting offering a voucher:
(11) Na09AV3A-1 (33: 08–33: 18)
01 CAR: che magari tu fai un buo:no (0.3) che sappiamo
that perhaps you make a voucher (0.3) so we know
02 suppergiù a quello che andiamo incontro e poi loro
more or less what we’re getting into and then they
03 possono fare quello che vogliono.
can do what they want
04 (0.6)
05 PIN: occhei >perché cioè n:el senso< ci siamo fatti
okay because that is in the sense we have made
06 più o meno i calcoli che tipo viene
more or less calculations that like it’s
07 duecentosessantano[ve a coppia,
two hundred and sixty-nine per couple
08 CAR: [brava.
good (girl)
In this case, Pina takes the turn at a TRP (l. 05) but displays difficulties in initiat-
ing her turn, hearable in her rapidly spoken >perché cioè n:el senso<. Whereas
cioè ‘that is’ has been described as a resource that speakers use to initiate self-
correction (e.g., Bazzanella, 1994, p. 24), this use has not been described, to my
knowledge, for nel senso (che). What has been observed, however, is the recurrence
of the collocation cioè nel senso in present-day conversational Italian (Fiorentini
& Sansò, 2017). However, whereas in Excerpts (6) to (10) nel senso (che) follows
a unit produced with falling intonation, in environments of self-repair it is also
observed as prosodically cohesive within a TCU-in-progress (Selting, 1996).
6. Turn-yielding
This section illustrates how participants trailoff (Jefferson, 1983; Local & Kelly,
1986) or yield (Auer, 1996; Ogden, 2001) a turn by ending their utterance with
the words nel senso (whereas nel senso che ‘in the sense that’ does not occur in this
position in my data). The following fragment is taken from a dinner table conver-
sation between five students living in Milan. They are discussing the price each of
them is asking for providing private lessons to high school students. Rino (RIN)
has just said that he is asking 50 euros per hour when Giorgio (GIO) claims that
some people ask 25 euros per hour (l. 01), which he appears to see as problematic
(ll. 04, 09). Angela (ANG) and Piera (PIE) are also sitting at the table.
42 Elwys De Stefani
(12) Mi13CE1–39 (07: 21–07: 35)

01 GIO: beh c’è gente che chiede venticinque all’ora eh?
well there’s people who ask 25 per hour right
02 (0.2)
03 RIN: eh: sì. (.) va beh. quel[lo (sì;si) m-
aw yes (.) well okay that (yes;one) m-
04 GIO: [(son) pezzi di m[erda
they’re pieces of shit
05 RIN: [dipende::
it depends
06 (0.2)
07 RIN: BAH:: dipe::nde io d- s-
well it depends I d- s-
08 [per venticinque ti #offro a]nche un&
at twenty-five I offer you also a
fig #fig.4
09 GIO: [da:i venticinque euro]
come on twenty-five euros
10 RIN: &*serv(h)i *zio dive:rso
different kind of service
RIN *.........*gazes at ANG-->
11 non so come +#di+re nel senso:.*+#
I don’t know how to say in the sense
RIN +...+right hand out-+
RIN -->*
fig #fig.5 #fig.6
12 *+(0.3)*
RIN *......*-->
RIN +right hand in-->
13 GIO: eh ma*+# che tanto non- non:: >cioè< le ripetizioni
well but anyway they don’t I mean the tutoring
RIN -->*gazes at GIO->>
RIN -->+
fig #fig.7
14 non garantiscono un cazzo °°(hai capito.)°°
lessons don’t guarantee fucking anything you know
The excerpt shows Giorgio and Rino exhibiting disagreeing opinions: Whereas for
Giorgio 25 euros appears to be too cheap a price for a tutoring lesson lasting an
hour, Rino holds that at that price he would offer a different service (i.e., involving
less preparation compared to a lesson for 50 euros per hour). Rino articulates his
position at lines 05–11 as disagreeing with Giorgio’s prior accusation that people
asking 25 euros are ‘pieces of shit’ (l. 04). His turn starts with dipende:: ‘it depends’
(l. 05), a further recurrently used projector phrase in Italian. He produces the core
of his statement as a complete clause (per venticinque ti offro anche un serv(h)izio

dive:rso ‘at twenty-five I offer you also a different kind of service’; ll. 08–10), after
which he utters the words non so come dire ‘I don’t know how to say’ (l. 11). His
turn ends with nel senso:. ‘in the sense’ produced with a lengthening of the final
syllable and a falling intonation. After a brief pause (l. 12), Giorgio takes the turn
and produces a potentially topic-closing statement (ll. 13–14).
Figure 4. Rino looking at his hand
Whereas the analysis based on talk alone could make us think that nel senso:. is
used here as a resource to organize the transition between Rino and Giorgio, a
look at the participant’s embodied conduct enables us to see that Rino is designing
his turn so as to allow other co-present participants to intervene at this point.
Figure 4 shows the participants’ orientation when Rino is in the midst of his turn
(l. 08). Shortly after, he will start looking at Angela, who is facing him, and he
holds this orientation until the end of his turn (Figure 5).
Figure 5. Rino looking at Angela
By gazing at Angela, Rino addresses her as an appropriate recipient of his talk

(see Goodwin, 1979), who might self-select as a possible next speaker (Stivers &
Rossano, 2010). Clearly, by ending his turn with nel senso:. Rino is not just yielding
the turn to the previous speaker (i.e., Giorgio), but he is actually opening up the
possibility for another speaker (i.e., Angela) to self-select, presumably in search
44 Elwys De Stefani
of an ally who might endorse his point of view. As Kangasharju (2002, p. 1448)
observed in committee meeting interactions, “[w]hen a disagreement between
two participants surfaces, an opportunity to join one of the parties is offered to the
other participants.” Rino’s turn-final nel senso:. occurring while he looks at Angela
offers precisely this opportunity.
But Rino’s nel senso:. is also concomitant with a further embodied phenom-
enon: As he pronounces the word dire (l. 11), he starts moving his right hand
outwards (which he holds half open above the table), and while he says nel senso:.
his hand reaches its rightmost position (Figure 6). This position is held until
Giorgio self-selects (l. 13), when Rino moves his hand towards resting position
again, while reorienting his gaze to Giorgio (Figure 7).
Figure 6. Rino moving left hand outwards
Figure 7. Rino gazes at Giorgio, hand resting
It is now possible to describe this use of nel senso:. as tightly related to Rino’s
concomitant embodied conduct – addressing with his gaze a different recipi-
ent (Angela) than the participant he is actually responding to and disagreeing
with (Giorgio), and performing an ‘opening’ gesture with his right hand. This
constellation of resources allows Rino to yield the turn and to make the floor avail-
able for a potential supporting party, which he guesses could be Angela.12
The following excerpt provides a further illustration of how turn-final nel
senso is coupled with a manual ‘opening’ gesture. We join the same group of per-
sons (now also in the presence of Giulio, the host) some 40 minutes later. Giorgio,
Piera, and Rino are discussing tidiness in flat-sharing communities: Piera has just
affirmed that she finds it important that the washbasin is kept clean, whereas Rino
has observed that the cleanliness of the washbasin does not matter very much
to him. It is in this environment of disagreement that Piera mentions a personal
experience of flat-sharing:
(13) Mi13CE1–39 (20: 59–21: 22) / Mi13CE2–45 (04: 34–04: 57)
01 GIO: io *che son-
I who am-
PIE *gazes at GIO-->
02 [io ti giuro io sono uno:
I swear I am someone:
03 PIE: [(bon) raga (.) anch’io +ho vissuto un anno+
(well) guys (.) me too I have lived a year
PIE +..................+
04 +per cazzi miei. [eravamo in-* i:: *eravamo+&
on my own we were we were
PIE +fingertips joined-------------------------+
PIE -->*,,,,,*gazes at GIO-->
05 GIO: [io s:-
I a:-
06 PIE: &+in +dieci. (.) in una casa [occhei?
ten (.) in a house okay
+...+hands open-->
07 GIO: [minchia.#
bugger me
fig #fig.8
08 RIN: eh:* quello:: non è u- quello::
er that that’s not a- but that
PIE -->*gaze not focused on any interactant-->
09 [è u- un'altra questione però.
is a- another question
10 PIE: [eravamo in dieci e:::# °°e-:°°
we were ten and e-
fig #fig.9
12. All of this is additionally facilitated by the fact that Giorgio is constantly looking away from
any potential recipient during this episode (Figures 4–7).
46 Elwys De Stefani
11 #nel senso:[:.#
in the sense
fig #fig.10 #fig.11
12 RIN: [vivere in die:ci
to live with ten people
13 #+è una cos*a::+# hippie.
is a hippie thing
PIE -->+.............+hands joined-->
PIE *gazes at RIN-->
fig #fig.12 #fig.13
14 (0.5)
15 RIN: non è vivere:[::
it’s not living
16 PIE: [macché hippie: ma che
no way hippie but what
17 caz[zo no-
the fuck no-
18 RIN: [che cazzo e- (cioè;c'è) era era pulita
what the fuck e- (that is) was was it clean
19 la casa in dieci persone?
the house with ten people
20 PIE: *no. (.) *infatti [la casa no.* ma il lavandino*&
no indeed the house not but the washbasin
PIE ->*,,,,,,,,*gazes at RIN--------*,,,,,,,,,,,,,,, *
21 GIU: [ah.
oh
22 PIE: &+*sì.*
yes
PIE -->+
PIE *gazes at RIN*
At lines 03–06, Piera relays what is recognizable as a personal experience of

her living on her own (i.e., separated from her family) with 10 people. Whereas
through her gaze Piera is selecting Giorgio as the addressee of her talk (Figure 8),
the latter provides only a minimal response, namely an expression of surprise (l.
07). Rino, instead, articulates a longer turn in which he treats Piera’s experience as
referring to ‘another question’ (ll. 08–09), thereby dismissing the relevance of her
talk for the topic at hand. While Rino addresses his talk to Piera, the latter does
not establish a reciprocal gaze with him, or with any other participant, as Figure 9,
from the perspective of the opposite camera, shows.
Figure 8. Piera gazes at Giorgio; joined fingertips
Figure 9. Piera’s unfocused gaze, hands opening
In overlap with Rino’s talk, Piera takes up her turn again and produces the words
eravamo in dieci ‘we were ten’ (l. 10) for the second time (see ll. 04–06). This
lexical recycling shows that she is resuming (Jefferson, 1972) her turn without
overtly responding to Rino’s criticism. However, her embodied conduct displays
the emergence of a momentary turn suspension, visible in her gaze – which is
wandering around without a clear focus – and in the unfolded hands, which reach
their maximal visibility when she utters the words nel senso::. ‘in the sense’ (l. 11;
Figures 10 and 11).
Figure 10. Piera further opening her hands

48 Elwys De Stefani
Figure 11. Maximal opening of Piera’s hands
Note that nel senso::. is pronounced with a lengthening of the final syllable and
with falling intonation – just as in Excerpt (12). Moreover, it is concomitant with
an ‘opening’ gesture of the hands, which reaches its apex at this moment. Clearly,
Rino hears and sees this conduct as opening up a possibility for turn-transition,
since he starts speaking again precisely towards the end of Piera’s talk, with mini-
mal overlap (l. 12). It is only in the course of Rino’s new turn that Piera orients her
gaze to him while again folding her hands, thereby displaying recipiency of Rino’s
talk, as Figures 12 and 13 show.
Figure 12. Piera touching left hand
Figure 13. Piera gazing at Rino, hands folded

In this section I have discussed two cases in which nel senso is involved in turn-
transition. I have observed prosodic (lengthening, downward intonation) as well
as embodied corollaries (‘opening’ gesture). Furthermore, in both excerpts, the
participants use other displays of a possible turn-end or suspension, such as Rino’s
non so come dire ‘I don’t know how to say’ in Excerpt (12), line 10 or Piera’s hearable
problems of turn-construction (e::: °°e-°°; Excerpt 13, l. 10). Evidence that partici-
pants orient to this specific use of nel senso can be seen in the facts that speakers
may end their turns with nel senso and allow a gap to occur, thereby making transi-
tion relevant (Excerpt 12), whereas hearers can treat the emergence of nel senso
as an occasion for identifying a TRP and hence for taking the turn (Excerpt 13).
A further characteristic of the turn-yielding use of nel senso pertains to its
prosodic embeddedness in the turn-in-progress. Indeed, in both excerpts, nel
senso is tied to the preceding talk in a prosodically cohesive way, since there is
no prosodic break prior to the articulation of nel senso. This makes this resource
similar to what Walker (2012) calls a ‘trail-off ’ conjunction.
7. Discussion and conclusion
The analysis has shown that speakers use nel senso (che) both for managing turn-
transition and for constructing turns-in-progress. The findings can be described
with respect to the following levels of analysis:
Turn-construction. Nel senso (che) is used:
a. at turn-beginning of a responsive turn (Excerpts 3–5);
b. at the beginning of a TCU of a turn-in-progress: using Schegloff ’s (2016)
terms, I have observed nel senso (che) in “next-beat” position (i.e., immedi-
ately after the end of the prior TCU; Excerpts 2, 7, 8), in “post-gap” position
(Excerpts 4, l. 05, Excerpt 9–10), and “post-other-talk” (Excerpt 6). In this re-
gard, Excerpt 4 (l. 09) shows an interesting case: The speaker uses nel senso che
in a responsive action, while neatly tying that action to her turn-in-progress.
This micro-sequential adjustment shows how sequential constraints may
impact turn-construction;
c. at turn-endings, where only nel senso (i.e., without che) occurs (Excerpts 12,
13); and
d. within an ongoing turn (Excerpts 11).
Syntax. When used in responsive turns, nel senso (che) introduces a new clause.
When used at TCU-beginnings, nel senso (che) enables speakers:
50 Elwys De Stefani
a. to combine clauses (Excerpts 2–8, 10);

b. to extend their turn with non-main clauses (see Excerpt 9, where the speaker
reintroduces a previously used lexical item, constituting the antecedent of a
relative clause).
Nel senso (che) can thus be described as a projector construction: whereas its use
as a conjunctive phrase is recognizable in (a), in (b) the projecting force of the
resource prevails.
Prosody. While it is impossible to make systematic claims on the basis of a
limited set of occurrences, three aspects can be highlighted:
a. in the TCU-initial use of nel senso (che), the resource follows an articulatory
break (Local & Kelly, 1986), which is hearable as a downward intonation on
the item preceding nel senso (che) (Excerpts 2, 7–10);
b. in that position, nel senso (che) may be followed by a pause (Excerpts 4, 6,
9), and is sometimes produced with falling intonation (Excerpts 4, 6). One
hypothesis, which has to be further investigated, is that this specific prosodic
and temporal formatting makes potentially expectable a more extended turn-
continuation. I found lengthening of the last syllable only in the format nel senso
(Excerpts 5, 7). These prosodic and temporal features show that the resource is
used by speakers and treated by listeners as a projector construction; and
c. I observed cohesive prosody (Selting, 1996) with items preceding nel senso
(i.e., without che) in self-repairs displaying turn-constructional problems
(Excerpt 11) and in its turn-yielding uses (Excerpts 12, 13). Only in the turn-
yielding occurrences, however, do the speakers produce this format with a
clearly downward intonation.
Action. In several of the excerpts analyzed, speakers orient towards some actual
or potential problem of understanding when using nel senso (che); this is where the
lexical meaning of senso ‘sense, meaning’ is still apparent:
a. responsive turns beginning with nel senso (che) occur after questions through
which speakers exhibit an actual problem of understanding (Excerpts 3, 4) or
verify correct understanding (Excerpt 5);
b. when used at TCU-beginnings, speakers orient towards a potential misun-
derstanding of their prior words. The possibility of a misunderstanding may
emerge from what speakers may treat as insufficient verbal (Excerpt 6) or em-
bodied (Excerpt 8) responsive actions by the recipient. In other cases, speakers
secure the correct understanding of a potentially problematic or imprecise
wording (Excerpts 2, 7, 9).
No issue of understanding is observed in Excerpt (10), where the speaker initiates

a confirmation request; the turn-yielding uses (Excerpts 12, 13) occur instead
in an environment of disagreement, but further research is needed to ascertain
whether this is a frequent praxeological environment.
Variation. The construction occurs in two formats, namely nel senso che and
nel senso. I found both manifestations in turn- and TCU-initial uses, whereas in
self-repair and turn-yielding, I observed only the format nel senso. The absence
of che in turn-yielding uses suggests that its occurrence would likely be heard as
projecting more talk.
Interactional research has examined a plethora of resources (formerly dis-
cussed under the heading of ‘discourse markers’) that speakers use to accomplish
tasks similar to what I have observed for nel senso (che). Bazzanella’s (1990, 1994,
2006) work has laid the foundation for the identification and description of such
resources in Italian (see also De Stefani, 2016 on niente ‘nothing’). Other languages
provide speakers with other resources, such as English I mean (Maynard, 2013),
Estonian tähendab ‘(it) means’ (Keevallik, 2003), Finnish mutta ‘but’ (Koivisto,
2010) or French je sais pas (Pekarek Doehler 2016), among many others, for which
both projecting and turn-yielding uses have been described. The fascinating par-
ticularity of nel senso (che) is that it occurs both as a projector construction – hence
projecting more talk by the same speaker – and as a trail-off device which, on the
contrary, makes speaker-change relevant. This chapter has offered a first investiga-
tion of nel senso (che) in present-day conversational Italian as a contribution to our
understanding of these formulaic chunks of talk.
References
Auer, P. (1996). On the prosody and syntax of turn-continuations. In E. Couper-Kuhlen, & M.

Selting (Eds.), Prosody in conversation (pp. 57–100). Cambridge, UK: Cambridge Univer-
sity Press. https://doi.org/10.1017/CBO9780511597862.004
Auer, P. (2009). On-line syntax: Thoughts on the temporality of spoken language. Language
Sciences 31, 1–13. https://doi.org/10.1016/j.langsci.2007.10.004
Bazzanella, C. (1990). Phatic connectives as interactional cues in contemporary spoken Italian.
Journal of Pragmatics 14, 629–647. https://doi.org/10.1016/0378-2166(90)90034-B
Bazzanella, C. (1994). Le facce del parlare: Un approccio pragmatico all’italiano parlato. Scan-
dicci: La Nuova Italia.
Bazzanella, C. (2006). Discourse markers in Italian: Towards a ‘compositional’ meaning. In K.
Fischer (Ed.), Approaches to discourse particles (pp. 449–464), Amsterdam: Elsevier.
Couper-Kuhlen, E. (2012). Turn continuation and clause combinations. Discourse Processes 49,
273–299. https://doi.org/10.1080/0163853X.2012.664111
interaction. Cambridge, UK: Cambridge University Press.
52 Elwys De Stefani
Dal Negro, S., & Fiorentini, I. (2014). Reformulation in bilingual speech: Italian cioè in German
and Ladin. Journal of Pragmatics 35, 1069–1093.
Dardano, M., Trifone, P. (1997). La nuova grammatica della lingua italiana. Bologna: Zanichelli.
Deppermann, A., & De Stefani, E. (2019). Defining in talk-in-interaction: Recipient-design
through negative definitional components. Journal of Pragmatics 140, 140–155.
De Stefani, E. (2016). ‘Niente’ nel parlato conversazionale: Pratiche interazionali e processi di
grammaticalizzazione di un segnale discorsivo. Zeitschrift für romanische Philologie 132(1),
206–231.
De Stefani, E. & Horlacher, A.-S. (2008). Topical and sequential backlinking in a French phone-
in program: Turn shapes and sequential placements. Pragmatics 18(3), 381–406.
https://doi.org/10.1075/prag.18.3.02ste
Fiorentini, I. & Sansò, A. (2017). Reformulation markers and their functions: Two case studies
from Italian. Journal of Pragmatics 120, 54–72.
Goodwin, C. (1979). The interactive construction of a sentence in natural conversation. In G.
Psathas (Ed.), Everyday language: Studies in ethnomethodology (pp. 97–121). New York:
Irvington Publishers.
Günthner, S. (2011). ‘N be that’-constructions in everyday German conversation: A reanalysis
of ‘die Sache ist/das Ding ist’ (‘the thing is’)-clauses as projector phrases. In R. Laury, & R.
Suzuki (Eds.), Subordination in conversation (pp. 11–36). Amsterdam: Benjamins.
https://doi.org/10.1075/slsi.24.03gun
Heritage, J. (1984). A change-of-state token and aspects of its sequential placement. In J. M.
Atkinson, & J. Heritage (Eds.), Structures of social action: Studies in conversation analysis
(pp. 299–345). Cambridge, UK: Cambridge University Press.
Jefferson, G. (1972). Side sequences. In D. Sudnow (Ed.), Studies in social interaction (pp.
294–338). New York: The Free Press.
Jefferson, G. (1983). On a failed hypothesis: ‘Conjunctionals’ as overlap-sensitive. Tilburg Papers
in Language and Literature 28, 1–33.
Jefferson, G. (2004). Glossary of transcript symbols with an introduction. In G. H. Lerner (Ed.),
Conversation analysis: Studies from the first generation (pp. 13–31). Amsterdam: Benjamins.
https://doi.org/10.1075/pbns.125.02jef
Kangasharju, H. (2002). Alignment in disagreement: Forming oppositional alliances in commit-
tee meetings. Journal of Pragmatics 34, 1447–1471.
https://doi.org/10.1016/S0378-2166(02)00073-5
Keevallik, L. (2003). From interaction to grammar: Estonian finite verb forms in conversation.
Uppsala: Acta Universitatis Upsaliensis.
Koivisto, A. (2010). Discourse patterns for turn-final conjunctions. Journal of Pragmatics 44,
Lee, Y.-A. (2006). Respecifying display questions: Interactional resources for language teaching.
Tesol Quarterly 40(4), 691–713. https://doi.org/10.2307/40264304
Lindström, J. (2006). Grammar in the service of interaction: Exploring turn organization in
Swedish. Research on Language and Social Interaction 39(1), 81–117.
https://doi.org/10.1207/s15327973rlsi3901_4
Local, J., & Kelly, J. (1986). Projection and ‘silences’: Notes on phonetic and conversational
structure. Human Studies 9(2/3), 185–204. https://doi.org/10.1007/BF00148126
Local, J., & Walker, G. (2004). Abrupt-joins as a resource for the production of multi-unit, multi-
action turns. Journal of Pragmatics 36, 1375–1403.
Maynard, D. W. (2013). Defensive mechanisms: I-mean prefaced utterances in complaint and
other conversational sequences. In M. Hayashi, G. Raymond, & J. Sidnell (Eds.), Con-
versational repair and human understanding (pp. 198–233). Cambridge, UK: Cambridge
University Press.
Mondada, L. (2016). Conventions for multimodal transcription. https://franzoesistik.philhist.
unibas.ch/fileadmin/user_upload/franzoesistik/mondada_multimodal_conventions.pdf
[Link accessed April 24, 2019].
bridge, UK: Cambridge University Press. https://doi.org/10.1017/CBO9780511620874
Ogden, R. (2001). Turn transition, creak and glottal stop in Finnish talk-in-interaction. Journal
of the International Phonetic Association 31(1), 139–152.
https://doi.org/10.1017/S0025100301001116
Pekarek Doehler, S. (2011). Clause-combining and the sequencing of actions: Projector con-
structions in French talk-in-interaction. In R. Laury, & R. Suzuki (Eds.), Subordination in
conversation (pp. 103–148). Amsterdam: Benjamins. https://doi.org/10.1075/slsi.24.06doe
Pekarek Doehler, S. (2016). More than an epistemic hedge: French je sais pas ‘I don’t know’ as
a resource for the sequential organization of turns and actions. Journal of Pragmatics 106,
Raymond, G. (2003). Grammar and social organization: Yes/no interrogatives and the structure
of responding. American Sociological Review 68(6), 939–967.
https://doi.org/10.2307/1519752
Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization of
turn-taking in conversation. Language 50(4), 696–735.
https://doi.org/10.1353/lan.1974.0010
Ochs, E. A. Schegloff, & S. A. Thompson (Eds.), Interaction and grammar (pp. 52–133).
Cambridge, UK: Cambridge University Press.
https://doi.org/10.1017/CBO9780511620874.002
Schegloff, E. A. (2016). Increments. In J. D. Robinson (Ed.), Accountability in social interaction
(pp. 239–263). Oxford, UK: Oxford University Press.
https://doi.org/10.1093/acprof:oso/9780190210557.003.0008
Selting, M. (1996). On the interplay of syntax and prosody in the constitution of turn-construc-
tional units and turns in conversation. Pragmatics 6(3), 371–388.
https://doi.org/10.1075/prag.6.3.06sel
Selting, M. (2005). Syntax and prosody as methods for the construction and identification of
turn-constructional units in conversation. In A. Hakulinen, & M. Selting (Eds.), Syntax
and lexis in conversation: Studies on the use of linguistic resources in talk-in-interaction (pp.
17–44). Amsterdam: Benjamins. https://doi.org/10.1075/sidag.17.04sel
Selting, M., & Couper-Kuhlen, E. (2001). Studies in interactional linguistics. Amsterdam: Ben-
jamins. https://doi.org/10.1075/sidag.10
Serianni, L. (1988). Grammatica italiana: Lingua comune e italiano letterario. Torino: UTET.
Stivers, T., & Rossano, G. (2010). Mobilizing response. Research on Language and Social Interac-
tion 43(1), 3–31. https://doi.org/10.1080/08351810903471258
54 Elwys De Stefani
Stoenica, I. M., & Pekarek Doehler, S. (this volume). Relative-clause increments and the man-
agement of reference: A multimodal analysis of French talk-in-interaction.
Thompson, S. A., & Couper-Kuhlen, E. (2005). The clause as a locus of grammar and interac-
tion. Discourse Studies 7, 481–505. https://doi.org/10.1177/1461445605054403
Walker, G. (2012). Coordination and interpretation of vocal and visible resources: ‘Trail-off ’
conjunctions. Language and Speech 55(1), 141–163.
https://doi.org/10.1177/0023830911428858
Zingarelli, N. (2018). Lo Zingarelli 2019: Vocabolario della lingua italiana. Bologna: Zanichelli.
Chapter 3
The emergence and routinization of

complex syntactic patterns formed with
ajatella ‘think’ and tietää ‘know’ in
Finnish talk-in-interaction
Ritva Laury and Marja-Liisa Helasvuo
Our paper concerns two Finnish cognitive verbs, ajatella ‘think’, and tietää
‘know’. We show that both verbs are most likely to occur in the first person
singular form but behave differently with respect to polarity: tietää occurs most
commonly in the negated form (56%), while ajatella is only rarely negated
(less than 4%). The verbs also differ with respect to their sequential emergence
and complementation, with tietää ‘to know’ occurring nearly half of the time
in responsive position and without complements. Each of the most common
formats of the verbs builds or projects a specific social action. The patterns
of clause combining, in this case, complementation or lack of it, are closely
connected to the locally contingent employment of action.
Keywords: cognitive verb, complementation, Finnish, fixedness
1. Introduction
Our article concerns patterns formed with two Finnish cognitive verbs, ajatella
‘think’, and tietää ‘know’, which are the two most frequently used cognitive verbs
and among the most frequently used verbs overall in the corpus we have used (see
Section 2 for details). We discuss the routinized patterns these verbs occur in,
the phonetic reduction they have undergone, the sequential emergence of these
patterns, and the social actions associated with them in everyday conversation. By
sequential emergence we mean the way the patterns are fitted to and shape their
sequential contexts. In our sequential analysis, we have focused on responsivity
and position within the turn, and how the patterns project subsequent talk.
https://doi.org/10.1075/slsi.32.03lau
56 Ritva Laury and Marja-Liisa Helasvuo
In earlier work, we have shown that these two verbs pattern very differently
with respect to various morphosyntactic criteria. Only certain forms are common,
and they are associated with particular social actions and behave differently with
respect to polarity and complementation (Helasvuo 2014a; Laury & Helasvuo 2016;
Laury, Helasvuo & Rauma forthcoming). As is common for verbs of cognition,1
both verbs are most likely to occur in the first-person form. However, they behave
differently with respect to polarity: tietää ‘know’ is most commonly negated, while
ajatella ‘think’ only rarely is.
Although both ajatella and tietää are verbs with full paradigms in terms of
the morphosyntactic markings of person, tense, and mood, they exhibit strongly
skewed patternings in actual language use. Only certain forms are frequent and
some are hardly used at all. Moreover, both verbs occur in certain high-frequency
expressions which show fixedness. This is again a feature that has been shown to be
typical of verbs of cognition with first-person subjects cross-linguistically.2
Both ajatella and tietää are normally thought of as complement-taking
predicates,3 but in our data, the verbs are used quite differently in this respect.
Ajatella is much more likely to be used with complements than tietää; in the
negative, tietää is used without a complement more often than with one. The verbs
also differ in terms of the type of complements they take: ajatella tends strongly
to occur with clausal complements, while the complement types tietää takes are
more varied. For both verbs, clausal complements are quite likely to come with
complementizers, for tietää even more commonly than ajatella.
We will show that both verbs occur in certain formulaic, semi-fixed expressions
which also differ in terms of their sequential emergence. For tietää, the expressions
are formed with the negative verb in the first person singular (henceforth, 1sg)
and the connegative form of the verb.4 As we will show, two different word orders
are possible and tend to have different sequential position. The negation-initial
1. See Helasvuo 2014a for Finnish; Scheibman 2002: 63 for English; Keevallik 2003 for Estonian;
Tao 1996: 25, 26, 124 for Mandarin; Travis 2007: 115–116 for Spanish.
2. For studies on ‘think’ and ‘know’ verbs in a range of languages, see e.g. Kärkkäinen 2003,
2012; Endo 2013; Deppermann & Reineke 2017, frthc.; Keevallik 2003, 2016; Maschler 2012,
2017; studies in Lindström, Maschler, & Pekarek Doehler 2016.
3. For complementation and complement-taking predications, see, e.g., Laury & Seppänen
2008; Laury & Helasvuo 2016 on Finnish; on German, Günthner 2011; on French, Pekarek
Doehler 2011, 2016; on English, Thompson 2002; Tao 2003; Thompson & Mulac 1991.
4. In Finnish, negation is formed with the negation auxiliary verb e-, which is marked for
person and number, but not for tense or mood. In negated clauses, the main verb comes in
the connegative form, which expresses tense and mood distinctions. For further discussion,
see Miestamo (2011).
Chapter 3. Emergence of syntactic patterns in Finnish talk-in-interaction 57
format tends to occur turn initially in responsive turns, without a complement,

while the pronoun-initial format is more likely to occur in a non-initial TCU
within a turn and to come with complements. In fact, initial negation and lack of
complement mark the first-person singular use of tietää as being responsive to the
prior turn. On the other hand, ajatella mostly occurs in 1sg past tense forms with
positive polarity and forms expressions which in our data functioned as framing
expressions which project further talk of a certain nature: expressions of stance,
the speaker’s future plans or her own thoughts.
The article shows that even within the same language, projective expressions
developed from cognitive verbs may come to have quite different profiles in terms
of what they project syntactically and in terms of social action. We suggest that
these disparate patterns reflect the use of these verbs in certain high frequency
expressions which show a great deal of formulaicity in particular sequential
contexts. In Section 3, we discuss the morphosyntactic profiles of the two verbs.
In Sections 3.2 and 3.3 we show, based on detailed microanalysis of the data,
how the most common patterns formed with these two verbs differ in their ac-
tion projection, sequential emergence and position within the turn. Section 4
contains a discussion of our findings contrasting the patterns formed with the
two verbs, and Section 5 presents our conclusions. We will first describe our
data and methodology.
2. Data and methodology
Our data come from Arkisyn, a morphosyntactically annotated corpus of everyday

conversations in Finnish (see Arkisyn 2018). These data have been supplemented
with additional data from the Recording Archive at the University of Turku, also
consisting of everyday conversation.5 Together the data amount to over 32 hours
of conversation.
In the Arkisyn corpus, tietää ‘know’ is the seventh most common verb with
1038 tokens. On the frequency list, ajatella ‘think’ is 17th among verbs (402
tokens). As noted, they are the two most frequently used cognitive verbs in the
corpus. Both ajatella and tietää favor 1sg, as 66.6% of the occurrences of ajatella
and 57.6% of tietää are in the 1sg form. Therefore, we are focusing on the use of
these verbs in that form. From our data, we have made two collections, one of
instances of ajatella ‘think’ (N = 337) and one of tietää ‘know’ (N = 619).
5. It cannot be claimed that our observations would apply to all of spoken Finnish; the use
of epistemic expressions may differ depending on genre (see, e.g., Maschler & Dori-Hacohen
2018).
The data were analyzed using the methodology of Interactional Linguistics

(Couper-Kuhlen & Selting 2018) with respect to syntactic, sequential, and phonetic
features.6 We coded for complementation or lack of it, and the syntactic category
of complement as clausal or other kind of complement (infinitival, nominal, or
other, such as pro-adverb). It turned out that especially the negation-initial format
of tietää ‘know’ (as in emmä tie ‘I don’t know’) occurs frequently without any
complement, while ajatella ‘think’ tends to occur with a complement (as in maat-
telin et mä voisin käydä juoksemassa ‘I thought that I could go for a run’).7 In the
case of clausal complements, we also coded whether there was a complementizer.
We have also studied the instances of the verbs within the larger sequential context
paying special attention to responsivity.
For both ajatella and tietää, certain high-frequency expressions dominate in
the data. We analyzed the most frequent patterns for features indicating phonetic
reduction (such as elision, merging of syllables, and cliticization of the subject
pronoun or complementizer). For ajatella, all first-person singular uses in our data
(248) were analyzed for phonetic reduction, while for tietää, we analyzed only a
subset (70 out of the total 620 uses in our data).
3. Complex syntactic patterns formed with ajatella ‘think’ and

tietää ‘know’
3.1 Morphosyntactic profiles
As we have indicated, ajatella and tietää show very different morphosyntactic

profiles in actual language use. Table 1 summarizes the differences between these
two verbs in our data.
As Table 1 shows, the verb ajatella ‘think’ strongly favors affirmative polarity
(only 2.4% were in the negative). Moreover, it favors past tense, especially in 1sg:
in our data, the overwhelming majority of 1sg uses of ajatella were in the past
tense (90.8%). In comparison, tietää is nearly the opposite of ajatella in terms of
polarity, especially when it comes to 1sg uses. As Table 1 shows, 85% of the 1sg
uses of tietää had negative polarity, while only 15% occurred in the affirmative.
Unlike ajatella, tietää favors present tense (95.2%).8
6. We are grateful to Ida Andersson, Mari Nikonen and Janica Rauma for their help in the
analysis and coding of the data.
7. See these expressions in their contexts in Examples 7 and 1, respectively.
8. Past tense forms represent 26.4% of all verbal predicates in the Arkisyn corpus (search made
November 13, 2018). The past tense of ajatella does not have the same pragmatic implications
Table 1. Morphosyntactic usage profiles of ajatella and tietää in the first person singular.
Tense Total Polarity
Present Past Other Affirm. Negative
Ajatella
(N = 337)
N 13 306 18 337 329 8
% of total 3.9% 90.8% 5.3% 100% 97.6% 2.4%
Tietää
(N = 619)
N 589 19 11 619 93 526
% of total 95.2% 3.1% 1.8% 100% 15.0% 85.0%
There is a clear morphological contrast between the present tense and past
tense in Finnish. It is not clear to us, however, what motivates the much more com-
mon use of the past tense in 1sgajatella in our corpus in many cases where there is
no evidence of the thinking having occurred in the past (see, e.g., Example 1, line
6). Further, in some of the most reduced formats for ajatella, the contrast between
present and past tense is neutralized (see e.g. Example 3, line 8).
In what follows, we will focus on the most entrenched patterns: past tense
forms of the verb ajatella ‘think’ in the affirmative (N = 248), and present tense
forms of the verb tietää ‘know’ with negative polarity (N = 503).
3.2 Ajatella ‘think’
In this section, we discuss the interactional functions of ajatella. We show that the
first-person singular form of this verb can be used to introduce expressions of the
speaker’s plans or epistemic or evaluative/affective stance. Furthermore, it can be
used in its literal meaning to introduce the speaker’s thoughts. That is, 1sg forms
of ajatella in our data are used in framing expressions which provide a point of
view or stance, or the like, for the clause which follows. As shown by Thompson
(2002), framing expressions are often formulaic, and in subsequent discourse, it is
the matter in the subordinate clause, and not the framing clause, which is focused
on by the participants in conversation.
In our data, ajatella is frequently used as a framing expression introducing
the speaker’s plans (39.5%). Studies on ‘I think’ expressions in other languages (cf.
of counterfactuality as the English I thought (Kärkkäinen 2012) or the German ich dachte
(Deppermann & Reineke 2017), possibly because Finnish has the verb luulla ‘presume’ which
carries these functions (Helasvuo 2014).
Kärkkäinen 2003, 2012, Deppermann & Reineke 2017, frth) do not mention this
function. Example 1 illustrates this. It comes from a phone conversation between
two friends who are discussing their plans for the day.
(1) SG112
01 VIK: nähdäänkö vielä il↑lalla.
see-pass-Q still night-ade
Are we going to get together tonight.
02 MIS: ↑joo, tulisik sää meille.h
ptc come-cond-Q 2sg 1pl-all
Yeah, would you like to come to my place.
03 VIK: voisim_ mä tulla.
can-cond-1sg 1sg come-inf
I could.
04 mää, meen Sallalle tässä. (0.2)
1sg go-1sg pn-all dem-ine
I will go to Salla’s place now.
05 heitän sen Pekkalaan jos se tarvii nii?,
throw-1sg dem-acc pn-ill if dem need-3sg ptc
I’ll drop her off at Pekkala if she needs to (go).
06 MIS: joo. .hh ↑kato ku maatteli e(t)
ptc ptc as 1sg-think-past comp
Okay. You see because I thought
07 mää voisin käydä ihaj juoksemassa?,
1sg can-cond-1sg go-inf even run-inf-ine
I could even go running
08 .hh nii tekiskö sum mieli tullam mukaa
ptc do-cond-Q 2sg-gen desire come-inf with
so would you like to come with (me)
09 vai tulisik sää sej jälkee,
or come-cond-Q 2sg dem-gen after
or would you (rather) come afterwards,
10 >ku mä lähen heti nyt<
because 1sg go-1sg immediately now
because I will go now right away
11 kolme(−-) jälkee.
three after
after three (−-).
12 (.)
13 MIS: .hhh
14 VIK: heti nyt,
immediately now
(You mean) right away.
15 (.)
16 MIS: nii.
ptc
Yeah
In Example (1), line 2, Missu first responds positively to Vikke’s suggestion that
they meet later the same evening (line 1) and presents an invitation to Vikke to
come to her house. Vikke accepts the invitation (line 3) and starts telling about
her other plans for the evening (lines 4–5). Missu responds to this with joo (line
6), which expresses understanding but at the same time indicates that the activity
(negotiating plans for tonight) is not yet complete (Sorjonen 2001: 48), and fol-
lows this with an account of her own plans (lines 6–7). Missu frames her account
with the mä ajattelin et ‘I thought that’ expression (line 6). Since she has issued
an invitation to Vikke to come to her place (line 2), her plans of not being home
are potentially problematic. In her study of workplace interactions, Stevanovic
(2013) suggests that constructing a proposal as a thought enables the symmetrical
distribution of deontic rights at the beginning of decision-making sequences.
Formulating her plan as a thought, Missu can be heard to frame as negotiable her
plan as well as the proposal that follows it. She continues with another invitation to
Vikke (lines 8–11) to join her run. The invitation is formulated as containing two
alternatives (to come with her for the run or to come later).
The proposal (lines 6–11) is syntactically a clause combination consisting of
two parts: The first part of the proposal is a ku(n) ‘when/ because’ adverbial clause
that contains maatteli e(t)9 ‘I thought that’ followed by a complement clause pre-
senting the plan (mää voisin käydä ihaj juoksemassa? ‘I could even go running’).
The second part is a nii ‘so’ main clause (lines 8–11) which contains the invitation
for Vikke to join the run. The first part (the kun-clause) provides the grounds for
the invitation presented in the second part. The construction in which mä ajattelin
et is housed is quite complex. Mä ajattelin et itself is embedded in an adverbial
clause, and is followed by a complement clause, and the adverbial clause is then
linked with nii ‘so’ to a clause combination, presenting two alternative plans. The
second alternative, linked with vai ‘or’ (line 9) to the first alternative, is followed
by another adverbial ku ‘when/because’ clause, which is rather loosely tied to
the whole semantically: it does not express the reason for what is expressed in
the clause that precedes it, nor is it a temporal expression tied to the preceding
clause. This shows, first of all, that the expression maatteli et does not contain the
main information subsequently discussed by the participants. It only provides an
epistemic frame, rather than functioning as a proper ‘main’ clause which would be
followed by a subordinate clause containing information that is less important or
subordinate to the information in the main clause. Here, the main information is
in the clause that follows it (cf. Thompson 2002). Secondly, while speakers are able
9. When we refer to the actual token, we use the actual (often quite reduced) form that appears
in that particular occasion of use. When we discuss the form in general, we use the fuller form,
mä ajattelin et.
to construct what at first glance seem highly complex clause combinations, parts
of such combinations may be only loosely linked together, and lack the semantic
and pragmatic values that their morphosyntactic characteristics indicate, as shown
with the last ku(n) ‘because/when’ clause in lines 10–11 (Laury & Ono 2014).
The data for ajatella show that the expression mä ajattelin et has many charac-
teristics of fixed expressions (see also Laury, Helasvuo & Rauma frthc., Wray 2002,
papers in Laury & Ono frthc.). It is very frequent in the data, and a close analysis
shows that over half (52%) of the clauses containing ajatella in the 1sg form occur
in this expression, i.e., they consist of the 1sg mä, the verb ajatella in the past
tense 1sg, the complementizer et(tä) and a complement clause. As noted above,
it is common for fixed expressions that they show signs of phonetic reduction
(see Bybee 2006, 2007, 2010). This is also true of the mä ajattelin et expression
in Example 1 (line 6): the pronominal subject is reduced and occurs as a clitic
pronoun attached to the verb (Helasvuo 2014b), the first two syllables of the verb
ajatella are merged into one (aa), the 1sg person marker -n is not produced, and
the complementizer että appears in a one-syllable form et. The transcription in
line 6 (Example 1) reflects the reduction.
In addition to its use in expressions of planning, ajatella commonly occurs in
contexts where it frames an expression of stance (epistemic, evaluative or affective;
see, e,g. Kärkkäinen 2003, Englebretson 2007) towards something (90/248; 36.3%).
Example 2 illustrates this. It comes from a conversation between a hairdresser and
her client. Prior to this, they have been discussing break-ins and car thefts. The
hairdresser (KAM) has told about the car of a mutual friend which had been stolen
but then found close to the harbor, and the police had said that it seemed to be
waiting there to be shipped away in order to be stripped for parts. The customer
(ASI) then tells how she decided against driving her car to the harbor when she
was leaving for a cruise.
(2) SG108
01 ASI: =.joo en mäkään uskaltanu ottaa autoo
ptc neg-1sg
1sg-clt dare-pcp take-inf car-par
I didn’t dare to take a car either
02 nyt ku me lähettii sinne risteilylle
now when 1pl go-pst there.to cruise-all
now when we went on the cruise
03 ku mä oisin jättäny sen sinne
as 1sg be-cond-1sg leave.pcp dem-acc there.to
I would have left it
04 rantaan, .hh vaikka siellähän on se parkki
waterfront-ill although there-clt be.3sg dem parking
at the harbor although there is that parking place
05 mut se#::#=mä en tiedä

but dem 1sg neg-1sg know-conneg
but I don’t know
06 onks se edes:(.) vartioitu.
be.3sg-q-clt dem even guard-pcp
if it’s even supervised.
07 hhh maatteli et se on kauhee työ
1sg-think-pst comp dem be.3sg awful job
I thought it would be an awful job
08 sit taas siihe (.).hh sitä ruveta hakemaan
then again dem-ill dem-par begin-inf search-inf
then again there- to begin to search for it.
09 ta#i:# [tai jos ] jotain ikkunoita rikotaan=
or or if some-par window-pl-par break-pass
or or if some windows get broken
10 KAM: [nii ja sit,]
ptc and then
Yeah and then
11 ASI:=tai m:uita nii.=
or other-pl-par ptc
or something.
12 KAM:=nii:,
ptc
Right.
In Example 2, having heard the hairdresser’s story about the mutual friend having
her car stolen, the customer tells about her own considerations when leaving for a
cruise: en mäkään uskaltanut ottaa autoo ‘I didn’t dare to take a car either’ (line 1).
She ponders the possible outcome had she taken the car with her (lines 3–9 and
11). Her description of a worst-case scenario is launched with the expression mä
ajattelin et (‘I thought that’), which functions as an evaluative frame. The evalu-
ation, here presented as a predicate nominal clause (line 7), se on kauhee työ ‘it
would be an awful job’ is syntactically built as its complement. In lines 8–9, she
describes one such scenario listing the inconveniences of having to start searching
for it, line 8, and having windows broken, line 9. She ends her three-part list with
the general extender (Overstreet 2014) tai muita (‘or something’, line 11; on list
construction, see Jefferson 1990, Selting 2007).
With respect to phonetic reduction, the mä ajattelin et expression in Example 2
(line 7) is similar to that in Example 1: the pronominal subject is cliticized to the
verb, the first two syllables of ajattelin are merged into one and form a syllable
together with the cliticized pronoun [maat], the final -n (the first person marker)
is dropped, and the complementizer is reduced to et.
The expression mä ajattelin että can also be used in its literal meaning, i.e. in
expressing one’s own thoughts. In her study of the English I thought, Kärkkäinen
(2012: 2199) found that I thought was rarely used to refer to a past cognitive act.
Deppermann and Reineke (2017) make a similar claim about the German ich
dachte expression.10 In the Finnish data, cases where the expression mä ajattelin
että was used to frame expressions of one’s own thoughts were not rare (22.8% of
the data). Example 3 illustrates this. It comes from the same conversation between
a hairdresser and her client as Example (2). Before this, the hairdresser has told the
customer how she fell on ice when she was walking her dog. Her story has come
to an end, and there is a lengthy pause. This is where the excerpt in Example (3)
begins, also about falling down on an icy street. It is occasioned by the hairdresser’s
story and offered as a second story (Sacks 1992: 769), relating a similar incident
evaluated from a similar perspective.
(3) SG108
01ASI: kyl mäki kaaduin ku: tota:
ptc 1sg-clt fall-pst-1sg as dem-par
I also fell down when
02 mä viel ajattelin just sillon ku.hh
1sg still think-pst-1sg just then when
I was even thinking right then when
03 mä menin sinne poliisiasemalle
1sg go-pst-1sg there.to police-station-all
I was going to the police station
04 ↑h(h)ak(h)een niitä rahoja niin,
fetch-inf dem.pl-par money-pl-par ptc
to fetch the money,
05 .hh mun edellä kaatu yks nainen
1sg-gen before fall.pst.3sg one woman
right in front of me a woman fell
06 ku oli iha (0.5) ↑iha vettä siellä. (0.5)
as be.pst.3sg fully fully water-par there
because there was actual water there
07 siinä jään päällä=
dem-ine ice-gen top-ade
there on top of the ice.
10. One of the editors suggests that register differences in the data might explain this difference
in the use of ‘think’ verbs in the three languages. Kärkkäinen’s (2012) English data, most of
which came from the Santa Barbara Corpus (Du Bois et al. 2000–2005), are quite comparable
to ours. Deppermann & Reineke’s German data came from the FOLK corpus, which contains
more types of conversational situations than ours. It is our judgment that the type of data used
cannot easily explain the differences in use patterns found.
08 KAM: =nii.=
ptc
Right.
09 ASI: =maattet voi ku mä en kaatuis
1sg-think-comp ptc as 1sg neg-1sg fall-cond-conneg
I thought I hope I won’t fall
10 .mt hh mut maattel(i) et se taksi on siinä
but 1sg-think-pst comp dem taxi be.3sg dem-ine
but I thought that the taxi is there ne-
11 £taksiautot siinä vieres et nyt ku
taxi.car-pl dem-ine next.to-ine comp now as
the taxi cars (are) there next to it so if
12 mä rojahdan tonne maahan niin
1sg slump.down-1sg there.to ground-ill ptc
I slump down there on the ground so
13 siinä sitä on. £.hh
dem-ine dem-par be.3sg
they are right there.
14 no enkös menny polvilleni. Mfhh.mt
ptc neg-1sg-q-clt go-pcp knee-pl-all-1sgpx
well didn’t I fall on my knees.
15 KAM: hmhh
16 ASI: maatteli että se(.)housunpolvi menee rikki
1sg-think-pst comp dem pants.knee go-3sg broken
I thought that the knee of my pants would be torn
17 mut ei se menny. (0.5) .hh
but neg.3sg dem go-pcp
but it wasn’t.
In Example (3), line 1 the customer starts to tell how she also fell. She ties her story
to the story she has just heard with the help of the clitic particle ki(n). The meaning
of this clitic particle is captured with ‘also’ in the free translation. In lines 2–4,
she focuses on what she had thought when she was walking to the police station.
A moment before she herself fell down, a woman who had been walking just in
front of her had fallen down on the icy surface with water on top of the ice (lines
5–7). The hairdresser shows recipiency with her response particle nii (line 8) at a
place of “maximum incompleteness” (see Sorjonen 2001: 233–238) and gives the
customer an opportunity to go on with her story. In line 9, the customer relates her
own thoughts at that moment, how she had hoped she would not fall. She reports
having thought about the taxis that were standing close by (lines 10–11) and about
how they would be available if she did fall down (lines 11–13). Line 14 contains the
climax of the story: despite her thinking that she should not fall, she did fall on her
knees. In line 16, there is yet another instance of the same expression reporting the
speaker’s thoughts, but at the same time, it shows the customer’s epistemic stance
(she had assumed the knees of her pants would be torn).
With regard to complementation, the expressions with ajatella in Example (3)
differ from each other: the first one (line 2) does not have a clausal complement
but is instead followed by a lengthy stretch of discourse containing a description
(lines 2–7) of where and when the incidents the customer is telling about took
place. In fact, in our data what follows mä ajattelin että is frequently not simply
a clausal complement; it is often a syntactically complex stretch of talk such as
the one here, consisting of several utterances rather loosely bound together (see
also Example 1, lines 7–11). (Cf. Thompson 2002; Hopper and Thompson 2008).
After the hairdresser’s response particle (line 8), the customer repeats the ajatella
expression with a cliticized complementizer et (line 9) which is followed by a
complement clause marked as a direct report of her thoughts with the particle
chain voi ku, the meaning of which is roughly captured by the free translation ‘I
hope’. The third one (line 10) is also followed by two complement clauses which
are both marked as complements with the complementizer et. The complement
clause in lines 10–11 reports the customer’s noticing the taxi cars close by and
the complement in lines 11–13 expresses her thoughts after she had noticed the
taxis. The final ajatella expression (line 16) also has a clausal complement pre-
ceded by the complementizer että. Example 3 illustrates the general patterns of
complementation of mä ajattelin että: the expression often occurs with a clausal
complement, but it may also be followed by a stretch of discourse whose connec-
tion with mä ajattelin että is looser (see also Example 1). In such cases, we can
no longer consider the use of the mä ajattelin että construction as part of a clause
combination, but rather as a “formulaic segment of speech that serves to project an
upcoming region of discourse” (Hopper & Thompson 2008: 119), which is neither
syntactically nor pragmatically subordinate to it.
The ajatella expressions in (3) show varying patterns of phonetic reduction:
the first one (line 2) shows hardly any reduction at all, but the others are more re-
duced. In all of them, the pronoun is cliticized to the finite verb as a mere [m] and
the first two syllables of ajatella ([a-ja]) are merged into one. In addition to this,
the medial l and the following i are not expressed (line 9), and the complementizer
että is reduced to et (lines 9 and 10). Looking at reduction in the whole corpus for
ajatella, it turns out that the most reduced form maattet (like the one in line 9) is
the most frequent pattern found in the data for expressions with ajatella in the
1st person singular: 16.5% (41 instances) of the instances exemplified exactly this
pattern (see Laury et al. frthc.).
Although the majority of instances of ajatella in 1sg consist of the sequence
mä ajattelin et(tä), there are quite a few cases where there are intervening elements
such as adverbs. Example (3) illustrates this: in line 2, there is an adverb viel in
between the pronominal subject mä and the verb ajattelin. Thus, in spite of its
tendency to appear in a fixed format, ajatella has in no way lost the syntagmatic
options available to this verb: the placement of the adverb viel ‘still’ in between the
subject and the finite verb follows normal word order patterns in Finnish.
In sum, ajatella is commonly used to frame expressions of the speaker’s plans,
stance, and her own thoughts. We have shown that in interaction, it is not the
framing expression but the framed expression that is focused on and forms the ba-
sis for subsequent talk. Despite its morphosyntactic potential, its actual use centers
around certain expressions. Most notably, mä ajattelin et(tä) + CLAUSE ‘I thought
that + CLAUSE’ is the most common pattern found in the data with over half
of the instances with ajatella exhibiting exactly this pattern, the 1sg form of the
verb preceded by a 1sg pronoun and followed by the complementizer et(tä) and
a clause. (See also Laury et al. frthc.). Syntactically, then, the 1sg past tense form
of ajatella can be said to project a clausal complement, although it can also initi-
ate longer stretches of discourse consisting of a number of clauses rather loosely
linked together. In its use, mä ajattelin et(tä) shows features typical of fixed expres-
sions, such as fixedness in morphological form (first person singular, past tense)
and loss of syntactic flexibility (fixed ordering of elements). Furthermore, it often
appears in a phonetically eroded form. Its elements do, however, show variation:
for example, the 1sg pronoun may appear as mä (as in Example 3, line 2) or mie
(as in mie aattelenki ‘I thought’), following the variation patterns typical of any 1sg
pronoun. This is, of course, a feature which is atypical of particles. Therefore, we
can say that it does not function as a particle, but rather, as a formulaic fragment
(Hopper & Thompson 2008). We have further shown that despite the tendency of
ajatella to fixedness, it still carries the syntagmatic options available to this verb.
3.3 Tietää ‘know’
In this section, we discuss the use of the first-person singular form of tietää. We
show that its sequential position and interactional function interact with its gram-
matical form. It is frequently used in responsive turns, where it is likely to occur
without a complement, and with a negation-initial word order. In general, the
format in which the negation verb is the initial element is likely to occur early in
the turn, while the other, pronoun-initial, word order more frequently occurs in a
non-initial TCU, and is followed by a complement. We also discuss the develop-
ment of certain expressions containing tietää into a particle.
Here is an example of the 1sg negated use of tietää in a responsive turn. The
participants in this conversation are three young women, gathered at the home
of one of them. Here they are discussing Anu’s new eyeglasses. Anu has told the
others that she had to return the glasses for repair the day after she purchased
them because one of the temples had broken and there was a scratch on one of
the lenses, so a new lens had to be ordered to replace the scratched one. There has
been a fairly lengthy discussion regarding the problems with the new glasses and
how they were occasioned.
(4) SG151
01 SUS: tuliks noihin nyt mitää muuta
come-pst-q-clt dem.pl-ill ptc any-par other-par
was there any other
02 muutosta siis #nii ku#,.hh näihi (.) ↑lins↑seihi.
change-par ptc ptc ptc dem.pl-ill lens-pl-ill
change to those like to these lenses
03 siis et tota joutuks ne
ptc comp ptc have.to.3sg-q-clt dem.pl
I mean like did they have to
04 korjaamaa lisää tai mitää,
repair-inf-ill more or any-par
do more repairs or anything
05 ANU: #e:m#mie tiiä.
neg-1sg-1sg know.conneg
I don’t know.
06 SUS: °joo°
ptc
Yeah.
07 ANU: kyl näil paremmi näkee ku niil.
ptc dem.pl-ade good-comp see.3sg ptc dem.pl-ade
you can see better with these than those.
08 (0.5)
09 SUS: joo
ptc
yeah
10 ANU: °toisil°
other-pl-ade
other ones
11 MIL: °hm°
12 ANU: °ei kai°
neg ptc
I guess not
13 SUS: °on siihe ehkä sit jotai pient° =
be.3sg dem-ill maybe then some-par small-par
there may be something small then
14 ↑eks se seli:ttäny et se laittaa siihe
neg.3sg-q-clt dem explain-pcp comp dem put-3sg dem-ill
didn’t (s)he explain that (s)he will put
15 jotai lisää sitä hioketta

something-par more dem-par grinding-par
something more on it that grinding (medium)
16 tai jo- (.) °emmie°,
or so- neg-1sg
or s- (.) I don’t,
17 ANU: prismaa ni[i]
prism-par ptc
prism, yeah
18 SUS: [n]i
ptc
yeah
19 ANU: ↑niin ↓kai emmie oi°kee tiiä°, (.)
ptc ptc neg-1sg-1sg really know-conneg
I suppose so I don’t really know,
20 em mie (°muista°)
neg-1sg 1sg
I don’t (°remember°)
In this example, Susa first asks Anu whether any other changes or repairs had to
be done on the lenses (1–4). The question is formulated in such a way, using the
indefinite pronoun mitää ‘anything’, that it projects an answer in the negative. In
line 5, Anu responds that she does not know; in other words, she denies epistemic
access to whether anything else was done to the lenses. Although an ‘I don’t know’
answer to a question fails to promote the agenda of the question, and is in that
sense uncooperative, an answer concerning a matter that the person the ques-
tion is addressed to has only indirect knowledge of (‘type-2 knowable’, Pomerantz
1980: 187) may not be treated as disaligning (Keevallik 2011a: 206), and indeed,
Susa receipts the response with a simple low-volume joo (line 6), expressing that
she has understood the content of the response, but possibly also implying that the
answer is not yet complete (Sorjonen 2001: 249; 48). Joo has also been argued to
occur in contexts where the participants have earlier discussed the matter at hand
(Sorjonen 2001: 251). This seems to be the case here.
Anu does go on to further discuss the matter, explaining that she can see bet-
ter with these glasses than niil ‘with those’, presumably her previous glasses (line
7), and after a short pause, she receives another joo response from Susa (line 9).
Although Susa’s response implies that she has understood the reference (Sorjonen
2001), Anu adds an increment, toisil ‘others’ (10), fitted as a head to the demonstra-
tive. She then provides another equivocal response to the earlier question, ei kai,
perhaps translatable as ‘I guess not; probably not’ (12). Susa takes a turn (lines
13–16) that indicates she has some knowledge of what had gone on at the optician’s
office. Her question is a negatively formatted interrogative beginning with eks se,
translatable as ‘didn’t s/he’. In English data, such interrogatives are thought to index
prior and independent access to the matter at hand and to convey the expectation of
a positive answer (Bolinger 1957: 99; Heritage 2002). An analogous strategy seems
to apply in Finnish too: Anu partly confirms this with her mention of prism (17),
and concedes the possibility with niin kai ‘I suppose so’, but then recycles her earlier
response, though in a somewhat hedged form, en mie oikee tiiä ‘I don’t really know’.
The form of both of Anu’s responses to Susa’s question are negation-initial.
This is the most common word order for 1sg uses of tietää in our data; nearly three
fourths of them (382/525, 72.6%) are negation-initial; the majority of the respon-
sive uses (302/382, 79.1%) consist of the initial negative verb in the 1sg form fol-
lowed by the first-person pronoun and the connegative form of tietää, in this order
(but as Example (4) line 19 shows, there may be intervening adverbs or particles).
In terms of phonetic reduction, note that the connegative verb form tiedä
occurs in the form tiiä, without the medial consonant, the initial [ie] diphthong
is produced as [i:] in line 5, and the nasal-final consonant of the negative verb
is assimilated to the initial nasal of the first person pronoun in terms of place of
articulation; word-final nasals regularly assimilate to the initial consonant of the
following word. The negative verb and the pronoun, em and mie are also pro-
nounced as one prosodic word.11 In our prosodic analysis of the subset of our data
of tietää (70 cases), there were 39 cases with verb-initial word order (i.e., negative
verb + pronominal subject + tietää in its connegative form, as in both instances
of this expression in Example (4)), and in almost all of these (36/39; 92.3%) the
negative verb and the pronominal subject formed one prosodic word. Further,
in our data, approximately half of the uses of negation-initial first-person uses of
tietää were assessed as phonetically reduced (19/39).
So far, we have seen that the first person singular negated form of tietää is
most commonly negation-initial and often reduced in form; nearly always, the
negative verb and its pronominal subject form one prosodic word. The major-
ity, nearly three fourths, of the ‘I don’t know’ responses in our data have the
negation-initial form.
Another feature of the responses in Example (4) is that they have no comple-
ments. This is of course to be expected, since the matter to which the speaker
is responding has just been mentioned in the preceding question and remains
intersubjectively accessible to the participants in the conversation. The negation-
initial uses of the first-person singular tietää in our data commonly come without
complements. Fewer than a third of them, 115/382 (30.1%) have complements.
11. In the identification of prosodic words, we followed Aho (2010), who has proposed the fol-
lowing as strong criteria for prosodic words: a prosodic word has a tonal rhythm pattern with
one stressed syllable and other non-stressed ones, and a coherent F0-contour. Additionally, we
used change in F0-contour and/or wave form and pause as weak criteria.
In fact, it has been suggested that the function of the negation-initial form
combined with the lack of a complement is to mark the turn as responsive to the
prior. In this sense, the format can be seen as emergent from the sequential position
of the turn; it occurs in second position, as a response to a preceding turn, and can
be said to both result from, and to function as a marker of, the responsive social
action it is accomplishing (see Hakulinen 2012; see also Lindström & Karlsson
2016 on negation-initial formats in Finland Swedish).
However, in many cases in our data, tietää in the negated form came with the
pronoun-initial word order (141/526; 26.8%). The pronoun-initial ‘I don’t know’
expressions are likely to come in the middle of or even late in the speaker’s turn
(99/141; 70.2%) and often have complements (97/141, 68.8%).
Even when the pronoun-initial 1sg negated tietää does not come in a non-
initial TCU, it is often still built as a continuation of the speaker’s turn. This is the
case in the next example. The example comes from a family dinner table conversa-
tion. Virpi, the mother, and Liisa, the daughter, have been discussing plans for the
next day. Liisa and her boyfriend Kasperi, also present, are planning to pick up his
parents after a short trip and drive them to their home. Liisa has just been telling
about those plans.
(5) SG441
01 LII: =mut ↑mikä se on sitte
but WH dem be.3sg then
but what is
02 se (0.2) huominen juttu °nyt.°
dem tomorrow-adj thing now
the thing tomorrow now then.
03 (0.2)
04 VIR: siis siel on (0.4) tänään ja huomenna (0.6)
ptc dem.loc-ade be.3sg today and tomorrow
Well today and tomorrow there’s
05 suklaa (.) festifaalit (0.6) vanhassa satamassa,
chocolate festival-pl old-ine port-ine
a chocolate festival in the Old Port
06 LII: m,
07 (1.2)
08 VIR: mut tota (.) mä en tiedä
but ptc 1sg neg-1sg know-conneg
but I don’t know
09 mi#tä siellä on#
WH-par there be.3sg
what (all) there is there
10 siel on varmaan #näyt#teilleasettajia

there be-3sg certainly exhibitor-pl-par
there are probably exhibitors
11 LII: mm.
12 (1.0)
13 VIR: sinne maksaa (.) ↑kakstoist euroo (0.8)
there.to cost-3sg twelve euro-par
it costs twelve euros (to get in)
14 [mut,
but
but,
15 LII: [°opiskelijalta ehkä vähemmän;°
student-abl maybe less
for a student maybe less;
In lines 1–2, Liisa asks her mother about an event taking place the next day, which
has apparently come up before; she refers to the event as se huominen juttu ‘the
thing tomorrow’, formatted with the definite determiner se (Laury 1997). Virpi
explains that there is a chocolate festival at Vanha Satama, a popular exhibit hall
(lines 4–5). After a pause, Virpi goes on to say that she does not know what all
there will be at the festival, but she assumes there will be exhibits (lines 8–10).
After a non-committal mm from Liisa (line 11), Virpi then mentions the price for
entrance to the festival, which is rather high, but Liisa proposes that there may be
a discounted price for students (lines 13–15). It has become evident earlier in the
conversation that Liisa is a student. Thus it seems that a possible visit by Liisa to
the festival is being negotiated.
In Example (5), the 1sg pronoun comes before the negative verb, and there
is a clausal complement. The utterance with mä en tiedä is built as a continuation
of Virpi’s preceding turn with the conjunction mut ‘but’ (line 8). Her turn may
be designed to downplay the importance of Liisa’s visit to the festival, and the
mention of the price may be working toward that end too. This may be motivated
by the fact that Virpi is giving a talk at the festival, and she is in a way giving her
daughter permission not to attend.
In our data, the pronoun-initial format of 1sg negated tietää (pro + NEG + V)
is more likely to be phonetically reduced than the negation-initial form (NEG + pro
+ V), although each contains the same elements, just in a different order. In
our prosodically analyzed sample, 18 of the 23 instances of the pronoun-initial
format were judged to be phonetically reduced, while approximately half of the
instances of the negation-initial format were, as noted above. This is somewhat
surprising, since the negation-initial format was more frequently used in our data
than the pronoun-initial format. On the other hand, in only 8 cases out of the 23
prosodically analyzed uses of this format, the pronoun and the negative verb were
separate prosodic words, while in the majority of the cases in the negation-initial
format, they were pronounced as one prosodic word. The greater tendency for the
pronoun-initial format to be reduced may be a consequence of its use as a projec-
tive framing device, an epistemic fragment. The fusion of the negative verb and the
pronoun in the negation-initial format may be due to the fact that the 1sg form of
the negative verb ends with a nasal and the first-person pronoun begins with one.
We have suggested that the pronoun-initial form of the first-person negated
form of tietää is more likely to come with complements than the negation-initial
form. It also tends strongly to occur in non-initial position in our data, and even
when it is in initial position, it is built as a continuation of the speaker’s turn, as
can be seen in Example (5). Even though this format occurs less frequently in our
data than the negation-initial form, it is more likely to be phonetically reduced.
In addition to the negative verb-initial and pronoun-initial formats discussed
above, another format for 1sg negated tietää in our data is one where the negative
verb comes first but there is no pronoun. This format appears to have developed
a particle-like use; in many of its uses, it no longer functions as a clause, and can
come at the end of the turn, or by itself in the middle of the turn. In these uses,
it functions as a hedge, indexing uncertainty on the part of the speaker. Consider
the next example. It comes from a conversation of two middle-aged colleagues
celebrating New Year’s together. At this point in the conversation, they are discuss-
ing a public hanging somewhere abroad that Kati had seen on television. Tarja has
asked whether the hanging had taken place in a closed-off space with crowds of
people. Kati responds to this:
(6) SG398
01 KAT: [#OO-OLI SIELLÄ >e< KATselijoita
be-pst dem.loc-ade onlooker-pl-par
there sure were onlookers
02 iha selkeesti mut oliko se sit
quite clear-adv but be.3sg-q dem ptc
quite clearly but was it
03 joku tori peräti vai oliks se vaan
some market.place actually or be-pst-q-clt dem only
actually a market place or was it just
04 sitte joku vankilan piha#: en tiiä
then some prison-gen yard neg-1sg know-conneg
some prison yard I dunno
05 (1.5)
06 KAT: mut katselijoita oli#:: [ihan
but onlooker-pl-par be.3sg-pst quite
but there were quite (a few) onlookers
07 TAR: [°mjoo:°
ptc
Ok:
08 KAT: paljon.
much
a lot.
In lines 1–4, Kati responds that there clearly were onlookers, but that she is not
sure what the location was. This part of her utterance is formatted as two polar
interrogatives (oliko se ‘was it’, line 2, vai oliks se ‘or was it’, line 3), which present
two guesses, marked as non-identifiable to herself as well as her recipient with
the indefinite determiner joku ‘some’ (Vilkuna 1992: 33): joku tori ‘some market
place’ and joku vankilan piha ‘some prison yard’. Since Tarja has clearly not seen
the program, and this turn is in fact an answer to Tarja’s question just prior, Kati is
not expecting an answer from Tarja. After a lengthened last syllable of piha ‘yard’
(line 4) uttered with creaky voice, indicating turn yielding (Ogden 2001), she adds
en tiiä. The placement of en tiiä here does not qualify it as a main clause of the two
question-formatted clauses. Indirect questions do not differ in form from polar
questions in Finnish, but their main clauses come before the indirect question.
Here, Kati’s en tiiä could be taken as an indication that she is uncertain about what
kind of place the location was, as an index of a non-committal stance between
the two guesses she presents (market place or prison yard). The placement of en
tiiä indicates that it does not take the prior clauses as complements, and it is thus
clearly functioning as an epistemic particle or fragment (Thompson 2002) rather
than a clause. Here, it may also function as a turn-exit device (cf. Pekarek Doehler
2016; Maschler 2017), together with the prosody.
In fact, both the negation-initial formats, with or without a pronoun, can be
used as nascent particles in our data. They both occur without complements in
contexts where they are not used responsively and when there is nothing in the
prior talk that the speaker of the negation-initial form could be denying epistemic
access to. Rather, they work as an expression of uncertainty, a hedge (cf. Weatherall
2011 on English and Maschler 2012: 2020–2021 for Hebrew analogues). Consider
the following example, taken from the same conversation as Example (4), showing
the use of the negative-verb initial format with a pronoun used as a hedge in a
non-responsive position. Anu has just told the others that she has purchased an air
purifier which is supposed to clean 99% of impurities from the air.
(7) SG151
09 ANU: et pitäs, (.) puhistaa kaikki nää (.) pölypunkit
comp should clean-inf all dem.pl dust.mite-pl
So (it) should clean out all of these dust mites
10 pois ja,
away and
and,
11 SUS: n[ii]
ptc
Yeah/I know
12 ANU: [huo]nepöly ja näi,
room.dust and dem.adv.mann
dust from the room and like this,
13 SUS: siite[pöly kaikki.]
pollen all
pollen and everything
14 ANU: [siitepöly ] ku mie [oon
pollen because 1sg be-1sg
pollen because I am
15 SUS: [.nff
16 ANU: niille kans kauheen allerkin[e,
dem.pl-all also terrible.adv allergic
terribly allergic to those too,
17 SUS: [°mm,°
18 ANU: emmie tie v:iime yön mie (.) vit:si
neg-1sg know-conneg last night-acc 1sg joke
I dunno last night I
19 kaks tuntii °h:ih° valvoin >ja< kuuntelin
two hour.par? stay.awake and listen-pst-1sg
stayed awake sheesh for two hours and listened to
20 sitä hurinaa siin, °miun piti
dem-par hum-par dem-ine 1sg-gen must-pst
that hum there, I had to
21 sam[muttaa se°]
turn.off-inf dem
turn it off
22 SUS: [kai siihe] tottuu. (.)
ptc dem-ill become.used.to-3sg
Probably (one) can become used to it.
In lines 9–17, Anu and Susa discuss the air purifier and its benefits. Both Anu and
Susa claim epistemic access to the qualities of the air purifier; see, for example,
Susa’s use of the particle nii(n) in line 11. Susa claims not just affiliation but in-
dependent access to the claims presented (Sorjonen 2001: 247). There is also a
considerable amount of overlap, indicating some epistemic competition; see, for
example, Anu’s early overlap line 14 (Vatanen 2014, 2018). Then, in line 18, Anu
digresses from the earlier direction of the air purifier discussion, initiating the
slight topic shift with emmie tie ‘I don’t know’. This expression consists of an initial
negative verb and a pronoun pronounced as one prosodic word, followed by the
connegative form of tietää. As is typical for the negation-initial uses of tietää, it

lacks a complement, but here it does not serve to deny epistemic access to any-
thing said prior to it, nor does it claim the speaker does not know what follows, a
complaint about having difficulty falling asleep due to the humming noise of the
device. We consider this a particle use; it is not responding to any prior claim, nor
does it serve to express not knowing what follows it. In several of the turn-initial
particle-like uses of this expression, it serves to mark a topic shift, what Keevallik
(2016) called a junction, as it does here, in addition to marking a shift in stance.12
A similar function has also been observed not only for Estonian (Keevallik 2016),
but also for French (Pekarek Doehler 2016) and Hebrew (Maschler 2012, 2017),
suggesting that there may be a cross-linguistic trend for ‘know’ verbs to mark
boundaries in conversation.
In sum, our data show that tietää is mostly negated in interaction. The nega-
tion-initial format containing a pronoun tends to be used responsively in our data,
and in responsive uses, there is most often no complement. The pronoun-initial
format is more likely to come with complements expressing what the speaker does
not know. Two of the formats, the pronounless negation-initial format, as well
as the negation-initial format containing a pronoun, also have particle-like uses,
which are not responsive to anything prior, and do not deny epistemic access to
anything following. Rather, they serve as hedges and may occur at boundaries or
junctions, where there is a shift in stance or topic.13 In our data, each format has
its own functions and sequential positions in which it occurs.
Next, we will compare the two verbs in our data, and discuss the implications
of our findings.
4. Discussion: Comparison and implications
As we have shown above, both ajatella and tietää are most likely to appear in the
first person singular form, as is typical for cognitive verbs, and both appear in
certain high-frequency, semi-fixed formats. However, the verbs differ in terms of
polarity and complementation, and have quite different functions, emerging in
distinct sequential contexts.
12. We thank Janica Rauma for her analysis of the turn-initial uses of the particle-like emmä
tiiä. – In Pekarek Doehler’s (2016) data, the French je sais pas also served as a turn-holding
device. There is some indication that emmä tiiä also functions in this way, but more data are
required to determine this.
13. It is possible that the different formats represent different stages of evolution of this expres-
sion, but space does not permit us to go into that aspect of the variation.
First-person singular forms of ajatella in our data are overwhelmingly af-

firmative; as shown in Table 1, only 2.4% are negated. Ajatella is used in the past
tense in the vast majority of cases in our data (90.8%). Thus the most frequently
occurring format of ajatella is the 1sg pronoun followed by the first person past
tense form of ajatella followed by the complementizer, mä ajattelin että ‘I thought
COMP’. Over half of the instances of ajatella (52%) follow this format (with no
intervening elements, e.g., adverbs or particles). This format is most often followed
by a clausal complement, or sometimes a lengthy stretch of discourse rather than
a single clause or clause combination. Thus, the format functions as a framing
expression that projects an expression of the speaker’s plan, stance, or the content
of her thought.
Like ‘think’ verbs cross-linguistically tend to, Finnish ajatella is frequently
used in expressions of epistemic and evaluative/affective stance. Differently from
‘think’ verbs in those languages where they have been studied (e.g. Kärkkäinen
2012 for English, Deppermann & Reineke 2017, frthc., for German), ajatella is
also used in planning contexts, to tell what the speaker is going to do; in our data,
this was its most common function. Also differently from what has been observed
for spoken English (Kärkkäinen 2012) and German (Deppermann & Reineke
2017; frthc.), it can be used to simply introduce expressions of the content of the
speaker’s thoughts.
The 1sg forms of tietää are clearly different from those of ajatella. In terms
of polarity, they tend strongly to be negated. Unlike ajatella, which is most com-
monly used in the past tense (90.8%), tietää strongly favors present tense (95.2%
of 1sg forms of tietää). Their most common use is in second position, i.e., in
responses to the prior turn, often to questions, but also to compliments, tasks of
identification, and other types of first pair parts anticipating responses. In this
position, the expression is most commonly negation-initial, and is not followed
by a complement. This follows from its sequential position: since the prior talk
contained the information that the speaker is now denying epistemic access to, the
negation-initial format and the lack of complement index the turn as a responsive
one. If the ‘I don’t know’ expression is pronoun-initial, it most often comes in the
middle of a speaker’s turn, often as a subsequent TCU in a turn, or if it is turn-
initial, that turn tends to be formatted as a continuation of the speaker’s prior turn.
In addition to the format with a pronoun, another format of the negated first-
person form of tietää is negation-initial but lacks a pronoun; this format appears
to be developing a use as an epistemic particle. It occurs in contexts where it does
not function to deny knowledge of anything particular, and in these uses, it tends
not to be followed by a complement, but rather indexes uncertainty. It could come
in the middle of a grammatical construction or even at the end of a turn. Those
uses where the expression lacks a complement and is not responsive, and could
not be denying epistemic access to anything in the prior talk, could be considered
particle-like; this would mean that 17.5% of the pronounless negation-initial for-
mat were particle-like. The negation-initial format which includes a pronoun also
has particle-like uses at only slightly lower frequency, 15.6%.
In general, in our data, tietää was most commonly used for epistemic negotia-
tion; one manifestation of this was the fact that after the 1sg form, the next most
frequent form was the second person singular, which appeared most commonly
in the (rather formulaic) interrogative form tietsä ‘do you know, you know’. Thus,
to generalize, tietää is used for asking about and denying epistemic access to
something. This is very different from the way ajatella is used in our data, which is
used to frame the expression of stance, plans, and the speaker’s internal thoughts.
It has been noted that for many bi-clausal expressions, what is normally
thought of the ‘main’ clause has become a formulaic fragment in everyday lan-
guage use which serves to project more talk of a particular nature to come (e.g.
Aijmer 2007; Hopper & Thompson 2008; Günthner 2011; Pekarek Doehler 2011;
Keevallik 2011b), and the matter in the subordinate clause is what the participants
focus on in the upcoming talk (Thompson 2002). This is often the case in our data
as well, especially when it comes to ajatella. What is central with the use of ajatella
is the plans, the stance, and the thoughts that are expressed. Tietää is different
in that while the content of the matter under epistemic negotiation is of course
important, just as important to the participants and the business at hand is the
status of the participants’ knowledge. This is reflected in the fact that tietää often
comes entirely without complements, while this rarely happens with ajatella.
Table 2 compares more closely the types of complements ajatella and tietää
take14.
Table 2 shows that in the majority of instances with tietää in the first person
singular, there is no complement (54.8%). If tietää does take a complement, it is
most likely to be clausal (29.7% of tietää had clausal complements). Ajatella has a
very different profile with respect to the complements it takes. For ajatella, comple-
ments are clausal in the majority of occurrences (73.4%). Both verbs may take a
nominal complement but it is not very common for either verb (5.2% for ajatella
and 10.5% for tietää). Infinitival complements are possible for both verbs but not
very common (7.3% of occurrences of ajatella had infinitival complements, but
there was only one case for tietää). The greatest difference between the two verbs
lies in their occurrence with no complement at all: it is much less common for
ajatella to have no complement at all than it is for tietää. This may be connected
14. In Table 2, “nominal + clause” refers to complements with a demonstrative or demonstrative

adjective head accompanied by an että clause (see Leino 1999; Vatanen, Suomalainen & Laury
frthc.) or, in the case of tietää, also an indirect question.
Table 2. Complement types for ajatella ‘think’ and tietää ‘know’ in the first person
singular.
Complement ajatella tietää
type N % N %
no complement 24 9.7 339 54.8
nominal 13 5.2 65 10.5
nominal + clause 5 2.0 13 2.1
clausal 182 73.4 184 29.7
infinitival 18 7.3 1 0.2
other 6 2.4 17 2.7
Total 248 100 619 100
to the fact that tietää, especially the negation-initial format emmä tiedä, as noted,
is commonly used in responsive position (61.7% of the verb-initial formats with a
pronoun are responsive) and without a complement (72.2% have no complement).
In contrast, the most common uses of ajatella are forward-linking: mä ajattelin että
functions as a projective frame to the speaker’s expressions of planning, stance, or
her own thoughts. Connected to its use as a framing expression is the fact that the
most frequent format for ajatella is found with the complementizer et(tä), which
is often cliticized to the verb, and a clausal complement, while the most frequent
formats of tietää lack a complementizer.
The very different profiles of the two most common Finnish cognitive verbs
may be connected to their semantic features (cf. Bybee & Pagliuca 1987). Ajatella
‘think’ is well suited for the expression of the speaker’s cognitive processes, be
it future plans, stance, or just plain internal thoughts. In contrast, tietää ‘know’
more directly involves epistemic matters. Talk about knowing something becomes
relevant when there is negotiation about whether and what someone knows, es-
pecially when there is lack of knowledge or uncertainty: thus it is commonly used
to deny the speaker’s epistemic access to, or to hedge an expression of knowledge
of something. They are so frequent in such uses that they become routinized
and reduced in form.
Both verbs occur in semi-fixed formulaic expressions. Over half (52%) of
the 1sg instances of ajatella represented the sequence given in (8) below, with no
intervening elements (such as adverbs). Similarly, there is one pattern for tietää
which is much more formulaic than others, namely that presented in (9). In our
data for 1sg instances of tietää, 35.3% of the data represented the sequence given
in (9) with no intervening elements.
(8) pro.sg1 + ajatella.pst.sg1 + et(tä) + CLAUSE

(9) neg.sg1 + pro.sg1 + tietää.conneg
The sequences given in 8 and 9 are semi-fixed in the sense that although the com-
bination of forms is as described, the phonetic outcome may vary. For example, the
form of the pronoun may vary: in the case of ajatella, it is often expressed as a clitic
pronoun attached to the verb (see Examples 1, 2, and lines 8 and 9 in Example 3),
but it may be an independent pronoun (as in line 2, Example 3). The pronoun may
also reflect patterns of areal variation (cf. mä in Example 5, but mie in 7).
Formulaic expressions often show phonetic reduction. Although tietää is used
more commonly in our database than ajatella, the most phonetically reduced 1sg
expression for these two verbs was a drastically reduced form of ajatella, namely,
maattet ‘I thought that’, in which the initial pronoun was reduced to a clitic
pronoun attached to the verb, the first two syllables of ajatella, [a] and [ja], were
merged to [a:], the medial [l] was dropped as well as the past tense marker i and
the complementizer että in reduced form et was cliticized to the verb. Interestingly,
this also was the most common phonetic form the 1sg instances of ajatella took in
our data: 16.5% of the cases with ajatella were of the form maattet, i.e. carried the
above-mentioned features of reduction. None of the 1sg forms of tietää were this
reduced. In fact, the most reduced format was not the most commonly used; the
pronoun-initial format in our data showed more reduction than the most common,
negation-initial format. This may be connected to its use as a projective frame.
Thompson (2002) has argued that formulaic complement-taking predications
do not function as proper main clauses, but are instead epistemic fragments. It
is possible that the high rate of reduction is connected to the frequent use of the
pronoun-initial format of tietää (mä en tiedä) and the mä ajattelin että -expression
as a projective expression. Of course, in the case of mä ajattelin että -expression,
it is not just used as an epistemic fragment, but rather, its functions as a projective
frame range from expressions of epistemic/evaluative/affective stance to framing
expressions of the speaker’s own thoughts or plans.
5. Conclusions
In this article, we have focused on the usage patterns of two cognitive verbs,
ajatella ‘think’ and tietää ‘know’ in Finnish conversational interaction. We have
shown that despite their full paradigmatic and syntagmatic potential, their use
is morphosyntactically strongly skewed. While both verbs have a strong prefer-
ence for first person singular, they show opposing profiles in terms of polarity and
tense. These preferences reflect the use of these verbs in certain high-frequency
expressions which show a great deal of formulaicity.
Both ajatella and tietää are complement-taking predicates. They may take
clausal or nominal complements, but as we have shown, clausal complements are
much more common for ajatella than for tietää. For ajatella, the most common
format covering over half of the instances in the data includes clausal comple-
ments. We concluded that this format functions as a framing expression, a for-
mulaic fragment (Hopper & Thompson 2008). What follows mä ajattelin että can
also be a lengthy stretch of discourse consisting of loosely connected utterances,
showing that the expression functions as a projective expression rather than a
main clause. The frame is followed by the expression of the speaker’s stance, plans,
or her own thoughts. For tietää, clausal complements are also quite common, but
importantly, over half of the instances with tietää, especially those in responsive
positions, had no complement at all. Two formats of tietää, the negation-initial
ones with or without the pronoun, show some signs of developing into a particle
in contexts where they are not used responsively. They can occur finally in a turn
or even in the middle of a grammatical construction.
Morphosyntactic erosion and phonetic reduction have been shown to be com-
mon for high-frequency fixed expressions (see e.g. Scheibman 2000, Kärkkäinen
2003, Keevallik 2003). We studied the data with respect to phonetic reduction, and
the analysis shows that reduction is common in first person singular expressions
with both verbs. For ajatella, reduction was very common as only 2 out of 248 in-
stances showed no features of reduction at all. It turned out that the most reduced
pattern (maattet) was the most frequent. For tietää, the format that showed the
most reduction was the pronoun-initial one. The majority of its uses in the data
we analyzed phonetically were judged to be reduced, which may be connected to
its use as a projective epistemic expression.
Our comparison of these two most frequently used Finnish cognitive verbs
showed that their grammatical formats are closely associated with and indeed
emerge from the use they are put to in interaction and from the sequential posi-
tions in which they occur.
Data source
Arkisyn. 2018. A morphosyntactically coded database of conversational Finnish. Database

compiled at the University of Turku, with material from the Conversation Analysis Archive
at the University of Helsinki and the Syntax Archive at the University of Turku. Department
of Finnish and Finno-Ugric Languages, University of Turku.
References
Aho, E. (2010). Spontaanin puheen prosodinen jaksottelu. [The prosodic sequencing of sponta-
neous speech.] Ph.D. Dissertation. Helsinki: University of Helsinki. https://helda.helsinki.
fi/bitstream/handle/10138/19251/spontaan.pdf?sequence=2
Aijmer, K. (2007). The interface between discourse and grammar. The fact is that. In A. Celle,
& R. Huart (Eds.), Connectives as discourse landmarks (pp. 31–46). Amsterdam: John
Benjamins. https://doi.org/10.1075/pbns.161.05aij
Bolinger, D. (1957). Interrogative structures of American English. Alabama: University of Ala-
bama Press.
Bybee, J. & W. Pagliuca, W. (1987). The evolution of future meaning. In A. Giacalone Ramat, O.
Carruba & G. Bernini (Eds.), Papers from the VIIth International Conference on Historical
Linguistics (pp. 109–122). Amsterdam: John Benjamins.
Bybee, Joan. (2006). From usage to grammar: the mind’s response to repetition. Language 82.4:
529–551. https://doi.org/10.1353/lan.2006.0186
Bybee, Joan. (2007). Frequency of use and the organization of language. Oxford: Oxford Univer-
sity Press. https://doi.org/10.1093/acprof:oso/9780195301571.001.0001
Bybee, Joan. (2010). Language, usage and cognition. Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511750526
Couper-Kuhlen, E., & Selting, M. (2018). Interactional linguistics. Cambridge: Cambridge
University Press.
Deppermann, A., & Reineke, S. (2017). Epistemische Praktiken und ihre feinen Unterschiede:
Verwendungen von ich dachte in gesprochener Sprache. In A. Deppermann, N. Proske, &
A. Zeschel (Eds.), Verben in interaktiven Kontext. Bewegungsverben und mentale Verben im
gesprochenen Deutsch. Tübingen: Narr.
Deppermann, A. & Reineke, S. Frthc. Practices of indexing discrepant assumptions with Ger-
man ich dachte (‘I thought’) in talk-in-interaction. In Functions of Language.
Du Bois, J. W. et al. (2000–2005). Santa Barbara Corpus of Spoken American English. Parts 1–4.
Philadelphia: Linguistics Data Consortium.
Endo, T. (2013). Epistemic stance in Mandarin conversation: The positions and functions of wo
juede (I feel/think). In Y. Pan, & D. Kádár (Eds.), Chinese discourse and interaction: Theory
and practice (pp. 12–34). London: Equinox.
Englebretson, R. (2007) Stancetaking in discourse: An introduction. In R. Englebretson (Ed.),
Stancetaking in discourse. Subjectivity, evaluation, interaction (pp. 1–25). Amsterdam:
Benjamins. https://doi.org/10.1075/pbns.164.02eng
Günthner, S. (2011). N be that-constructions in everyday German conversation: a reanalysis
of die Sache ist/das Ding ist (‘the thing is’)-clauses as projector phrases. In R. Laury, &
R. Suzuki (Eds.), Subordination in conversation: A cross-linguistic perspective (pp.11–36).
Amsterdam: John Benjamins. https://doi.org/10.1075/slsi.24.03gun
Hakulinen, A. (2012). Näkökulmia suomen kieltolauseen sanajärjestyksen määräytymiseen.
[Perspectives on word order in the Finnish negated clause]. In R. Argus, R. Hussar, & T.
Rüütmaa (Eds.), Pühendusteos emeriitprofessor Mati Hindi 75. sünnipäevaks [Festschrift for
the 75th birthday of Prof.Emer. Mati Hint] (pp. 20–44). Tallinna Ülikooli eesti keele ja
kulttuuri instituudi toimetised 14. Tallinn: Tallinn University Press.
Hakulinen, A., Vilkuna, M., Korhonen, R., Koivisto, V., Heinonen, T., & Alho, I. (2004). Iso
suomen kielioppi [The Comprehensive Grammar of Finnish]. Helsinki: Suomalaisen Kirjal-
lisuuden Seura.
Helasvuo, M.-L. (2014a). Agreement or crystallization: Patterns of 1st and 2nd person subjects
and verbs of cognition in Finnish conversational interaction. Journal of Pragmatics 63,
Helasvuo, M.-L. (2014b). Searching for motivations for grammatical patternings. Pragmatics
24(3), 453–476. https://doi.org/10.1075/prag.24.3.02hel
Heritage, J. (2002). The limits of questioning: Negative interrogatives and hostile question con-
tent. Journal of Pragmatics 34, 1427–1446. https://doi.org/10.1016/S0378-2166(02)00072-3
Hopper, P. J., & Thompson, S. A. (2008). Projectability and clause combining in interaction. In
R. Laury (Ed.), Crosslinguistic studies of clause combining: The multifunctionality of conjunc-
tions (pp. 99–124). Amsterdam: John Benjamins. https://doi.org/10.1075/tsl.80.06hop
Jefferson, G. (1990). List construction as a task and a resource. In G. Psathas (Ed.), Interaction
competence (pp. 63–92). New York: Irvington.
Kärkkäinen, E. (2003). Epistemic stance in English conversation. Amsterdam: John Benjamins.
https://doi.org/10.1075/pbns.115
Kärkkäinen, E. (2012). I thought it was very interesting. Conversational formats for taking
stance. Journal of Pragmatics 44(15), 2194–2210.
Keevallik, L. (2003). From interaction to grammar. Estonian finite verb forms in conversation.
Acta Universitatis Upsaliensis. Studia Uralica Upsaliensia 34. Uppsala.
Keevallik, L. (2011a). The terms of not knowing. In T. Stivers, L. Mondada, & J. Steensig (Eds.),
The morality of knowledge in conversation (pp. 184–206). Cambridge: Cambridge University
Press. https://doi.org/10.1017/CBO9780511921674.009
Keevallik, L. (2011b). Interrogative “complements” and question design in Estonian. In R. Laury,
& R. Suzuki (Eds.), Subordination in conversation: A cross-linguistic perspective (pp.37–68).
Amsterdam: John Benjamins. https://doi.org/10.1075/slsi.24.04kee
Keevallik, L. (2016). Abandoning dead ends: the Estonian junction marker maitea ‘I don’t know’.
Keevallik, L., & Hakulinen, A. (2018). Epistemically reinforced kyl(lä)/küll-responses in Esto-
nian and Finnish: word order and social action. Journal of Pragmatics 123, 121–138.
Laury, R. (1997). Demonstratives in interaction: The emergence of a definite article in Finnish.
Amsterdam: John Benjamins. https://doi.org/10.1075/sidag.7
Laury, R., and Seppänen, E.-L. (2008). Clause combining, interaction, evidentiality, participa-
tion structure, and the conjunction-particle continuum: The Finnish että. In R. Laury (ed.),
Crosslinguistic Studies of Clause Combining. Amsterdam: Benjamins.
https://doi.org/10.1075/tsl.80.08lau
Laury, R., & Helasvuo, M.-L. (2016). Disclaiming epistemic access with ‘know’ and ‘remember’
in Finnish. Journal of Pragmatics 106, 80–96. https://doi.org/10.1016/j.pragma.2016.07.005
Laury, R., Helasvuo, M.-L., & Rauma, J. Frthc. Use of the verb ajatella ‘think’ as a fixed expres-
sion in spoken Finnish. Submitted.
Laury, R., & Ono, T. (2010). Recursion in conversation: What speakers of Finnish and Japanese
know how to do. In: H. van der Hulst (ed.), Recursion and human language. Berlin: Mouton
De Gruyter. Pp. 69–92. https://doi.org/10.1515/9783110219258.69
Laury, R., & Ono, T. (2014). The limits of grammar: Clause combining in Finnish and Japanese
conversation. Pragmatics 24(3), 561–592. https://doi.org/10.1075/prag.24.3.06lau
Lindström, J., & Karlsson, S. (2016). Tensions in the epistemic domain and claims of no-
knowledge: A study of Swedish medical interaction. Journal of Pragmatics 106, 126–147.
Lindström, J., Maschler, Y., & Pekarek Doehler, S. (Eds.) (2016). Grammar and negative epis-
temics in talk-in-interaction: Cross-linguistic studies. Special issue of Journal of Pragmat-
ics, 106.
Maschler, Y. (2017). The emergence of Hebrew loydea/loydat (‘I dunno masc/fem’) from inter-
action. In A. Sansò, & C. Fedriani (Eds.), Pragmatic markers, discourse markers and modal
particles: New perspectives (pp. 37–69). Amsterdam/Philadelphia: John Benjamins.
Maschler, Y., & Dori-Hacohen, G. (2018). Constructing a genre: Hebrew lo yode’a / lo yoda’at ‘(I)
don’t know’ on Israeli political radio phoneins. Text & Talk 38(5), 575–604.
Miestamo, M. (2011). A typological perspective on negation in Finnish dialects. Nordic Journal
of Linguistics 34(2), 83–104. https://doi.org/10.1017/S0332586511000126
Ogden, R. (2001). Turn transition, creak and glottal stop in Finnish talk-in-interaction. Journal
of the International Phonetic Association 31(1), 139–152.
https://doi.org/10.1017/S0025100301001116
Overstreet, M. (2014). The role of pragmatic function in the grammaticalization of English
general extenders. Pragmatics 24:1.105–129.
Pekarek Doehler, S. (2011). Clause-combining and the sequencing of actions: projector
constructions in French conversation. In R. Laury, & R. Suzuki (Eds.), Subordination in
conversation: A cross-linguistic perspective (pp. 103–148). Amsterdam: John Benjamins.
Pekarek Doehler, S. (2016). More than an epistemic hedge: je sais pas ‘I don’t know’ as a resource
for the sequential organization of turns and actions. Journal of Pragmatics 106, 148–162.
https://doi.org/10.1016/j.pragma.2016.06.014.
Pomerantz, A. (1980). Telling my side: “limited access” as a fishing device. Sociological Inquiry
50: 3–4. 186–198.
Sacks, H. (1992). Lectures on conversation. Ed. by Gail Jefferson. Oxford: Blackwell.
Scheibman, J. (2000). I dunno. A usage-based account of the reduction of don’t in American
English. Journal of Pragmatics 32(1), 105–124.
https://doi.org/10.1016/S0378-2166(99)00032-6
Scheibman, J. (2002). Point of view and grammar: Structural patterns of subjectivity in American
English conversation. Amsterdam: John Benjamins. https://doi.org/10.1075/sidag.11
Selting, M. (2007). Lists as embedded structures and the prosody of list construction as an
interactional resource. Journal of Pragmatics 39(3), 483–526.
Sorjonen, M.-L. (2001). Responding in conversation: A study of response particles in Finnish.
Amsterdam: John Benjamins. https://doi.org/10.1075/pbns.70
Stevanovic, M. (2013). Constructing a proposal as a thought: A way to manage problems in
the initiation of joint decision-making in Finnish workplace interaction. Pragmatics 23(3),
519–544 . https://doi.org/10.1075/prag.23.3.07ste
Tao, H. (1996). Units in Mandarin conversation: Prosody, discourse, and grammar. Amsterdam:
John Benjamins. https://doi.org/10.1075/sidag.5
Tao, H. (2003). A usage-based approach to argument structure: ‘Remember’ and ‘forget’ in
spoken English. International Journal of Corpus Linguistics 8(1), 75–95.
https://doi.org/10.1075/ijcl.8.1.04tao
Thompson, S. A. (2002). “Object complements” and conversation: Towards a realistic account.
Thompson, S. A., & Mulac, A. (1991). A quantitative perspective on the grammaticalization of
epistemic parentheticals in English. In P. J. Hopper, & E. C. Traugott (Eds.), Approaches to
grammaticalization, Vol. 2 (pp. 313–339). Amsterdam: John Benjamins.
Travis, C. (2007). Genre effects on subject expression in Spanish: Priming in narrative and
conversation. Language Variation and Change 19, 101–135.
https://doi.org/10.1017/S0954394507070081
Vatanen, A. (2014). Responding in overlap. Agency, epistemicity and social action in interac-
tion. Ph.D. dissertation, Department of Finnish, Finno-Ugrian and Scandinavian Studies.
Helsinki: University of Helsinki.
Vatanen, A. (2018). Responding in early overlap: Recognitional onsets in assertion sequences.
Research on Language and Social Interaction 51(2), 107–126. (https://doi.org/10.1080/083
51813.2018.1413894)
Vatanen, A., Suomalainen, K., & Laury, R. Frthc. The Finnish projector phrase se että as a fixed
expression.
Vilkuna, M. (1984). Voiko -kin-partikkelia ymmärtää? [Can the particle -kin be understood?]
Virittäjä 88: 393–408.
Vilkuna, M. (1992). Referenssi ja määräisyys suomenkielisten tekstien tulkinnassa. [Reference
and definiteness in the interpretation of Finnish texts]. Helsinki: Suomalaisen Kirjallisu-
uden Seura.
Weatherall, A. (2011). I don’t know as a prepositioned epistemic hedge. Research on Language
and Social Interaction 44(4): 317–337. https://doi.org/10.1080/08351813.2011.619310
Wray, A. (2002). Formulaic language and the lexicon. Cambridge: Cambridge University Press.
Chapter 4
The insubordinate – subordinate continuum

Prosody, embodied action, and the emergence of
Hebrew complex syntax
Yael Maschler
This chapter examines the continuum stretching between Hebrew syntactically

integrated and unintegrated (‘insubordinate’) she-‘that/which/who’-clauses
produced following final/continuing intonation contours in naturally-occurring
interaction. Leaving aside modal insubordinate she-clauses, I show that in all of
these cases she- ties back to an immediately prior stretch of interaction – verbal
and/or embodied – and projects an elaboration or evaluation of it, without
much concern about which particular type of complex construction – relative,
complement, or adverbial (if any) – is being created. The data suggest that rather
than viewing insubordinate clauses as imperfect realizations of the canonical
‘subordinate’ variety resulting from the disintegration of complex syntactic
patterns, canonical, syntactically integrated varieties of Hebrew relative,
complement, and adverbial clauses may be regarded as grammaticizations from
syntactically less integrated varieties.
Keywords: subordination, insubordination, Hebrew, grammaticization, prosody
1. Introduction1
Traditional approaches to syntax (e.g., Lyons, 1968; Blau, 1966) stipulate that any
noun phrase (NP) or adjectival phrase (AdjP) in a clause may be substituted by a
clause. In the first case, a complement or adverbial subordinate clause is concerned;
in the second, a subordinate relative clause. Clauses are thus viewed as embedded
within ‘higher’ clauses. Languages are shown to exhibit various morphosyntactic
1. I thank Jan Lindström, Simona Pekarek Doehler, Paul Hopper, Sandy Thompson, and Betty
Couper-Kuhlen for their constructive comments on earlier versions of this chapter. This study
was supported by the Israel Science Foundation, grants no. 887/12 and 1233/16 to Yael Maschler.
https://doi.org/10.1075/slsi.32.04mas
88 Yael Maschler
means for marking the embedded status, or the structural dependency, of the
subordinate clause on the matrix clause. However, it has long been noted that
embeddedness and dependency do not always go together; thus the concepts of
‘cosubordination’ (clauses that are [+dependent, -embedded], Olson, 1981; Van
Valin, 1984: pp. 548–549) and ‘insubordination’ (Evans, 2007) were introduced.
In a recent study, Evans and Watanabe modified Evans’ study (2007), adding
that insubordination can be viewed “diachronically as the recruitment of main
clause structures from subordinate structures, or synchronically as the indepen-
dent use of constructions exhibiting prima facie characteristics of subordinate
clauses” (2016: p. 2).
So-called ‘subordinate’ clauses bearing loose or no syntactic relations to an
element in a preceding clause have received some attention recently in the spoken
discourse of several languages2. I add here an interactional linguistic (Selting &
Couper-Kuhlen 2001; Couper-Kuhlen & Selting, 2018) perspective on this issue,
investigating spoken Hebrew discourse from a multimodal perspective.
In previous approaches to insubordination, insubordinate clauses are viewed
as imperfect realizations of the canonical ‘subordinate’ variety resulting from
the disintegration of the patterns of complex syntax. For instance, in her cross-
linguistic survey of insubordination studies, Cristofaro notes that according to
those studies “[t]he relevant clauses acquire a self-standing status” (Cristofaro,
2016: p. 419, emphasis mine) for one of three reasons: (1) ellipsis of the matrix
clause (e.g., Evans, 2007), (2) extension of dependency “from sentence-level syn-
tax into larger discourse and pragmatic domains” (Mithun, 2008: p. 69), or (3)
clausal disengagement – [d]ue to highly particularized contextual circumstances,
the linkage between a subordinate […] clause and an accompanying main clause
is weakened, until the subordinate clause can be used independently” (Cristofaro,
2016: pp. 402). The findings of the present study support the opposite view. Rather
than viewing insubordinate clauses as imperfect realizations of the canonical ‘sub-
ordinate’ variety resulting from the disintegration of complex syntactic patterns,
I argue that, because the two varieties implement essentially the same discourse
actions, syntactically integrated ‘subordinate’ clauses may be regarded as gram-
maticizations from syntactically less integrated varieties. In other words, I suggest
that the canonical, syntactically integrated varieties of Hebrew relative, comple-
ment, and adverbial clauses arise out of looser sequences of clauses.
The clauses examined in the present study are of two kinds: (1) syntacti-
cally unintegrated by virtue of not being embedded in any matrix clause, or
2. See, e.g., Lehti-Eklund, 2002; Anward, 2004; Evans, 2007; Mithun, 2008; Laury & Seppänen,
2008; Keevallik, 2008; Verstraete, D’Hertefelt & Van Linden, 2012; Weinert, 2012; Mertzlufft &
Wide, 2013; Günthner, 2014; Wide, 2014; all studies in Evans & Watanabe, 2016; Inbar, 2016.
Chapter 4. The insubordinate – subordinate continuum 89
(2) syntactically loosely-integrated, in the sense that they cannot be classified

unambiguously as one particular type of subordinate clause (relative, complement,
or adverbial) because some morphosyntactic dependency marker3 is ‘missing’. Yet
both syntactically unintegrated and syntactically loosely-integrated types mani-
fest one prima facie characteristic of subordinate clauses (Evans and Watanabe,
2016: p. 2); namely, they all begin with she- ‘that/which/who’ – an element known
as the “‘unmarked’ all-purpose [subordinating] conjunction” in traditional
Modern Hebrew grammar (Glinert, 1989: pp. 308–310, cf. Blau, 1966: p. 93). They
are therefore both termed here ‘insubordinate’.
The etymology of Hebrew she- has been debated in a number of studies. Most
scholars believe it to be derived from the Akkadian relativizer ša-, itself derived
from a masculine singular accusative demonstrative pronoun in Old Akkadian
(Bergsträsser, 1909; Holmstedt, 2007; but see Huehnergard, 2006; Deutscher,
2009). Synchronically, according to traditional Modern Hebrew grammar, the
morpheme she-, which attaches (at least orthographically)4 to the first word of
the following clause, is the most frequent Hebrew ‘subordinator’ introducing the
subordinate clause in bi-clausal relative5 and complement clauses. The conjunc-
tion she- also combines with other components to form a variety of adverbial
clauses, forming various adverbial conjunctions such as kshe- ‘when/while’, lifney
she- ‘before’, mipney she- ‘because’, kedey she- ‘in order that’, bemikre she- ‘if ’,
lamrot she- ‘although’, among many others. Finally, she- begins the subordinate
clause constituting part B of pseudo-cleft constructions (Hopper, 2001; Hopper
& Thompson, 2008), and it attaches to the question words ma ‘what’ and mi ‘who’
forming nominalized clauses.
However, recent studies “taking temporality seriously” (see Hopper’s
Afterword to this volume; Hopper, 1987, 1992, 2011; Deppermann & Günthner,
2015; Pekarek Doehler et al., 2015), show that this traditional description of
Hebrew subordination does not account for many complex clause constructions
found in spoken Hebrew discourse (Maschler, 2011, 2012, 2017, 2018; Maschler &
Nir, 2014; Polak-Yitzkhaki & Maschler, 2016; Maschler & Fishman, forthcoming).
3. Other than she- (see below).
4. There are in fact many instances throughout the database in which the morpheme she- is
separated from the first word of the following clause by an intonation unit boundary (Chafe,
1994; Du Bois et al., 1992; Du Bois, forthcoming. See, e.g., Excerpt 6, lines 87 88 below).
5. In the case of relative clauses, syntactic integration of the subordinate clause is marked not
only by the subordinator but also by an obligatory resumptive pronoun in the relative clause
which agrees in gender and number with the NP antecedent in the matrix clause. According
to traditional Hebrew grammar, only accusative resumptive pronouns may be ‘dropped’ (e.g.,
Glinert 1989: pp. 361–362; Tsadka, 1989).
90 Yael Maschler
In particular, based on an audio-recorded corpus, I have shown that there are many
she-clauses that cannot be clearly classified into any of these ‘subordinate’ catego-
ries and argued that syntactically un- or loosely-integrated she-clauses are better
understood in terms of the actions they implement in discourse (Maschler, 2018).
I have argued that this conception of un- or loosely-integrated she- clauses sheds
light also on the syntactically integrated cases and suggested positioning syntacti-
cally integrated and syntactically unintegrated clauses at two ends of a continuum
of integratedness, depending on the lexical properties of the linguistic element
preceding the she-clause (Maschler, 2018: pp. 699–701, cf. Fox & Thompson,
2007). The present study examines a related continuum of integratedness stretch-
ing between syntactically integrated and unintegrated she-clauses – this time
based on the morphosyntactic means for marking the dependency, and adding a
multimodal analysis of some of the data – thus yielding new findings.
Following presentation of the corpus (Section 2), I move to an analysis of all
insubordinate she-clauses found throughout the database. The clauses were first
classified according to their general discourse function – elaborative, modal, or
evaluative. In the present study I provide an interactional linguistic analysis of the
evaluative (Section 3) and elaborative (Section 4) varieties, leaving aside the modal
type.6 In Section 5 I conclude the study and discuss its implications.
2. Data and overview
The study is based on over 11 hours (669 minutes) from the Haifa Multimodal
Corpus of Spoken Hebrew (Maschler et al., 2019), of casual conversation between
students, their friends and relatives, audio-recorded between the years 1993–2014,
and fully transcribed and segmented into intonation units (Chafe, 1994; Du Bois
et al., 1992; Du Bois, forthcoming). The corpus included at the time of data col-
lection 243 conversations among 701 speakers, 2–5 per interaction. The data
also includes video-recordings of 9 interactions of similar genre, recorded dur-
ing 2017–2018, a total of approximately 2 hours of talk among 22 interlocutors,
2–4 per interaction.
The audio corpus reveals roughly 200 tokens of syntactically un- or loosely-
integrated she-clauses; the video corpus manifests 33 – altogether approximately
235 instances in 13 hours of talk. The phenomenon is thus far from rare.
The present study is based on a close sequential analysis of 121 of the audio
tokens and of all 33 instances found in the video recordings – altogether 154
6. For the modal variety, see Maschler (2018: pp. 673–674, 683).
tokens. All syntactically un- or loosely-integrated she-clauses in the database were

classified according to their general discourse function:
Table 1 shows that the majority of these clauses function elaboratively (98,
64%), next in frequency come the modal clauses (41, 26%), and finally the evalua-
tive ones (15, 10%).7 For the sake of ease of presentation, we begin with the latter.
Table 1. Distribution of insubordinate she-clauses according to discourse function

Function N
Elaborative 98 (64%)
Modal 41 (26%)
Evaluative 15 (10%)
Total 154 (100%)
3. Evaluative insubordinate she-clauses
Approximately 10% (N = 15) of all insubordinate she-clauses throughout the data

provide the speaker’s stance concerning a previous utterance. In the great majority
of cases (N = 13) an evaluative stance is at issue.8 Evaluative insubordinate she-
clauses are found in the corpus usually turn-medially (14 out of 15 tokens, 93%),
but they may also occur turn-initially.
Excerpt (1) comes from a conversation between two young Israeli men
discussing the Chinese language. Alex tells Dotan that there is a transliteration
system of Chinese in the Latin alphabet ('alefbet 'angli ‘English alphabet’, as he calls
it), in which Chinese tones (which he calls 'intonatsyot ‘intonations’) are indicated
by diacritics (for which he uses the diminutive chupchikim ‘little strokes’) over
the Latin letters:9
7. As is generally the case with linguistic categorizations, occasionally we find an insubordinate

she-clause showing that the boundaries between these categories may be fuzzy. Nevertheless, a
classification is viable, since even in the ambiguous cases it is possible to argue for the dominance
of one category over the others. For an example of a fuzzy boundary between the elaborative and
evaluative categories, see Maschler (2018: pp. 693–696).
8. In an additional two cases epistemic stance is conveyed.
9. Transcription conventions are based on Chafe (1994), Du Bois et al. (1992), and Du Bois
(forthcoming) with a few changes (see Appendix). Transcription of embodied behavior is based
on Mondada (2018).
92 Yael Maschler
(1) ‘Chinese Poetry’

100 ALE: ...naxon be-sinit yesh ta--'intonatsyot?
right in-Chinese there_is OBJ.DEF-intonation.PL
...right in Chinese there are the intonations?
101 DOT: ...naxon.

right.
...right.
102 ALE: ...'az-- 'e--h,

...so-- u--h,
103 'ata mesamen 'otam 'im--,

you.M.SG indicate.PRS.M.SG OBJ-3M.PL with
you indicate them by--,
104 chupchikim,
little_stroke.PL
little strokes,
105 le-kol miney kivunim.

to-all kind.PL.GEN direction.PL
in all kinds of directions.
106 DOT: be-'emet?

in-truth
really?
107 ALE: ..ken.
..yes.
108 DOT: ...'aval sinit,

...but Chinese,
109 ...ze safa she.. ke'ilu--,

this.M.SG language that as_if
...is a language that.. li--ke,
110 ...ze lo rak 'intonatsyot,

this.M.SG NEG only intonation.PL
...it’s not only intonations,
111 ze--,
this.M.SG
i--t’s,
112 ....ke'ilu
as_if
....like
113 ...mamash 'ex

really how
...really how
114 ..'ex she-'ata--,

how that-you.M.SG
..how you--,
{dim}
115 ALE: lo,

NEG
no,
116 'aval ma she-ha-..'alefbet ha-sini ma'avir,

but what that-DEF alphabet DEF-Chinese get_across.PRS.M.SG
but what the.. Chinese alphabet gets across,
117 she-ha-'alefbet ha-'angli lo ma'avir,

that-DEF-alphabet DEF-English NEG get_across.PRS.M.SG
that the English alphabet does not get across,
118 hu rak ha-'intonatsyot.

he only DEF-intonation.PL
is only the intonations.
119 DOT: ...m.
120 ...mhm.
...mhm.
{creaky voice}
121 ...me'anyen.
...interesting.
122 ALE: she-ze dey pashut.

that-this.M.SG pretty simple
which is pretty simple.
122 ke'ilu kav... ka-ze ze--.. flet?,

as_if line like-this.M.SG this.M.SG flat
like a line like this i--s flat?
123 ...ma--,
...ma--,
(continues illustration, see ex. 8 below)
Following Alex’s telling (1. 100–105), Dotan responds with the admirative dis-
course marker be'emet? ‘really?’(1. 106) (Maschler & Estlein, 2008: pp. 292–294),
but then begins to object (1. 108–114). Alex interrupts this with lo, 'aval ma she-ha-
..'alefbet ha-sini ma'avir, she-ha-'alefbet ha-'angli lo ma'avir, hu rak ha-'intonatsyot.
‘no, but what the Chinese alphabet gets across, that the English alphabet does not
get across, is only the intonations’ (1. 115–118), ending in a sentence-final intona-
tion contour (1. 118).
While the creaky voice at the end of intonation unit 118 prevents us from
seeing the falling pitch contour in the Praat (Boersma & Weenink, 2019) spectro-
94 Yael Maschler
graph (Figure 1), we can clearly see the declining pitch and intensity along the four
intonation units 115–118 leading to this final intonation (Figure 2): 10
File Edit Query View Select Spectrum Pitch Intensity Formant Pulses Help
135.957224 1.066420 (0.938 / s) 137.023643

0.1668
Ch 1
0
-0.1254
0.1668
Ch 2
0
-0.1254
5000 HZ 500 HZ
555.5 HZ 112.8 HZ
0 HZ 75 HZ
0.090 1.066420 1.432080
135.866822 135.866822 Visible part 2.588901 seconds 138.455723 524.302916
Total duration 662.758639 seconds
ze rak ha-’intonatsyot
is only the intonations
Figure 1. ‘Chinese Poetry’, line 118.
132.930292 4.057484 (0.246 / s) 136.987777
0.4294
0 Ch 1
-0.4529
0.4294
0 Ch 2
-0.4529
5000 HZ 500 HZ
1328 HZ
142.4 HZ
0 HZ 75 HZ
0.658285 4.057484 0.462033
132.272008 132.272008 Visible part 5.177802 seconds 137.449809 525.308830
Total duration 662.758639 seconds
Figure 2. ‘Chinese Poetry’, lines 115–118
10. The only ‘hump’ (Matalon, 2016) towards the end of this series of intonation units is at lo
‘no’ of line 117 and has to do with the contrastiveness expressed in intonation units 116-117 (cf.
Chafe, 1994: pp. 76–78).
Dotan accepts Alex’s assertion with m …mhm and the evaluation me'anyen ‘inter-
esting’ (1. 119–121), and Alex overlaps the final syllable of this assessment with an
evaluation of his own.
Alex’s evaluation she-ze dey pashut. ‘which is pretty simple’ (1. 122) is thus first
of all prosodically unintegrated into any other clause. It is also syntactically unem-
bedded in any matrix clause: it cannot be considered a relative clause according to
traditional syntax, because there is no AdjP which this clause may be seen to sub-
stitute. Furthermore, a Hebrew relative clause must exhibit a resumptive pronoun
coreferential with some NP antecedent in the matrix clause (unless the correferent
is in the accusative), but no such NP antecedent is to be found. The demonstrative
ze (1. 122) – the only candidate for resumptive pronounness here – refers back
to Alex’s previously-raised topic of the transliteration method of Chinese tones
in the Latin alphabet, specifically to 'ata mesamen 'otam 'im--, chupchikim, le-kol
miney kivunim. ‘you indicate them by--, little strokes, in all kinds of directions.’
(1. 103–105) which he subsequently refers to metonymically by ha-'intonatsiyot
‘the intonations’ (1. 118).11 The she-clause at (1. 122) is thus unembedded in any
matrix clause and therefore classified as insubordinate. It is [+dependent, -embed-
ded] and can also be considered ‘cosubordinate’ (Van Valin, 1984).
This she-clause functions to provide the speaker’s evaluative stance concerning
prior discourse, in this case the straightforwardness of the transliteration method
in his judgement. Indeed, in his following utterances (1. 122–123 and Excerpt 8
below), Alex proceeds to illustrate the simplicity of the method.
Line 122 is produced as an increment (Schegloff, 1996; Ford, Fox & Thompson
2002; Auer, 2007; Couper-Kuhlen & Ono, 2007; Linell, 2013) of the post-other talk
variety (Couper-Kuhlen & Ono, 2007: p. 514; Schegloff, 2001), because on the one
hand, it is produced as a syntactic continuation of Alex’s previous clause (which
had ended in a complex transition relevance place (1. 118), i.e., in a point of pro-
sodic, syntactic, and pragmatic completion (Ford & Thompson, 1996)), but on the
other hand, it follows Dotan’s affiliative response (1. 119–120) and the greater part
of his assessment (1. 121).
We are reminded of Clift’s study of ‘non-restrictive which-increments’ in
English conversation, in which she argues that such increments may function
“to assess or evaluate an aspect of that expressed in the main clause” in pursuit of a
display of common stance (Clift, 2007: pp. 55–56, see also Tao & McCarthy 2001).
Later on Dotan receipts Alex’s illustration of the system with 'ah. 'okey. ‘oh. okay’
displaying understanding of Alex’s description (8. 139–140 below). However, as
11. Note that ha-'intonatsiyot ‘the intonations’ cannot be the antecedent; it is not only semanti-
cally incongruent, but it is also a feminine plural noun and therefore does not agree with the
masculine singular demonstrative ze (1. 122) as a ‘proper’ Hebrew antecedent should.
96 Yael Maschler
we have seen, there is no ‘main’ clause here to which Alex’s insubordinate she-
clause provides the evaluative stance.
Furthermore, not all evaluative insubordinate she-clauses in the data are
produced as increments: 7 out of 15 (47%) of the tokens follow continuing, rather
than final intonation in same-speaker talk, as in the following excerpt from a
conversation between two close friends, Eyal and Aya:
(2) ‘Ammonia Leak’
26 EYA: 'a,
oh,
27 'etmol 'e--h,
yesterday u--h,
28 dibarti 'im kati,

speak.PST.1SG with Kati
I spoke with Kati,
Eyal Aya
fig. 3 3
fig.
29 she-ze--,
that-this.M.SG
which i--s,
fig. 4
30 tsk
31 AYA: xadash?
new?
32 EYA: /?/
33 *(1.57)
eya: *-->smiling-->
34 ken,
yeah,
eya: -->--
35 AYA: nadir?
rare?
eya: -->--
36 EYA: ze haya nexmad,

this.M.SG be.PST.3M.SG nice
it was nice,
eya: -->---------------------------
37 AYA: @
38 EYA: 'ulay namshix @la'asot @'et @ze.*

maybe continue.FUT.1PL do.INF ACC this.M.SG
maybe we’ll continue doing it.
eya: -->-----------------------------------------*
39 AYA: @@@@@
fig. 5
Eyal begins to tell his friend Aya the story of how he asked his partner Kati to
move in together: 'etmol 'e--h, dibarti 'im kati, she-ze--, tsk ‘yesterday u--h, I spoke
with Kati, which i--s, tsk’ (2. 27–29). The more-to-come slightly rising intona-
tion contour at (2. 28) can be clearly seen in the pitch trace on the final syllable
of ‘Kati’ (Figure 6).
Eyal’s utterance ends in continuing intonation with a sound stretch
(she-ze-- ‘which i--s’) and is followed by a dental click (transcribed as tsk) (2.
29–30), reminiscent of Wright’s description of English clicks in word searches
(2005: pp. 176–228).12 While Eyal’s gaze has been on Aya throughout line 28
(Figure 3), he averts his gaze to the side at the beginning of the she-clause, touch-
ing his chin with his right hand (Figure 4), displaying a thinking face (Goodwin
& Goodwin, 1986) and suggesting that he is cognitively searching for a word. Aya
responds to this verbal, vocal, and embodied behavior offering a continuation of
12. This is actually somewhat different from the English word searches involving clicks in-
vestigated by Wright, as in her data recipients never collaborate in the word search (Wright
2005: p. 227).
98 Yael Maschler
0.1847
-0.2729
0 2.7
Time (s)
250
200
150
Pitch (HZ)
100
50
’a ’et mol ’e-- di bar ti ’im ka ti she ze--
oh yesterday uh I spoke with Kati which is
0 2.7
Time (s)
Figure 6. ‘Ammonia Leak’, lines 26–29
the clause she-ze-- ‘which i--s’ begun at (2. 29): xadash? ‘new?’ in appeal intonation
(Du Bois, forthcoming).13 This constitutes a humorous completion evaluating the
prior clause (2. 27–28) (clearly, it is not a new matter that Eyal and his partner
talk to each other), a stance which Eyal’s utterance, gaze and gesture up to this
point did not necessarily convey (although there is a hint of a smile on Eyal’s face
already at line 28 (Figure 3), which his close friend may have detected). In any
event, Eyal now smiles more recognizably and responds affirmatively with ken
‘yeah’ (2. 33–34), and Aya offers an upgraded completion: nadir? ‘rare?’ (2. 35).
To this Eyal, still smiling, and turning his gaze back to Aya, responds with the
assessment ze haya nexmad ‘it was nice’ and a humorous upgrade of his own: 'ulay
namshix la'asot 'et ze ‘maybe we’ll continue doing it’ (2. 36, 38). These lines are
accompanied by the laughter of both participants (2: 36–39, Figure 5), attesting to
the humorous stance project these two friends are mutually engaged in.
13. As Du Bois writes (forthcoming, §5.3, pp. 1–2), intonation contours constitute linguistic
signs, and just like any other linguistic sign, comprise a signifier and a signified (de Saussure,
1959[1916]). In the case of appeal intonation, the signifier is an intonation contour ending in a
high rise in pitch, and the signified in English (as well as in Hebrew) is the very general meaning
“‘I’m seeking an (implicit) response from you’” (Du Bois, forthcoming, §5.3: pp. 1–2, see also Du
Bois et al., 1992: pp. 30–31).
The co-constructed (Lerner 1991) she-clause she-ze--, xadash? ‘which i--s,

new?’ (2. 29, 31), with its upgrade nadir? ‘rare?’ (2. 35), again, cannot be seen to
substitute for any AdjP in a matrix clause and is therefore unembedded in any
matrix clause, yet it is not produced as an increment. This is the case with about
half the evaluative insubordinate she-clauses throughout the data.
Eleven more insubordinate she- clauses function evaluatively throughout the
corpus.
4. Elaborative insubordinate she-clauses
Much more frequent are the elaborative insubordinate she-clauses (98 tokens,
64%), a function which was found for such clauses also in Dutch, German, Sweden-
Swedish, Finland-Swedish, and Italian14. I have shown that Hebrew elaborative
insubordinate she-clauses emerge online as speakers are performing a variety of
tasks and responding to local interactional contingencies (Maschler, 2018), such
as negotiating a delicate moment in interaction, aligning with an interlocutor,
attributing upcoming discourse to another speaker (cf. Keevallik, 2008; Laury &
Seppänen, 2008), and formulating the orientation to a narrative (Labov, 1972).
However, a meta-level action is common to all these tokens:
she- is employed as a ‘wildcard’ tying back to an immediately prior stretch of
interaction and projecting [Hopper, 2001; Auer, 2005; Günthner, 2011; Pekarek
Doehler, 2011; Maschler, 2012, 2015] an elaboration (a specification or explica-
tion) […] of it, without worrying too much about which particular type of
complex construction – relative, complement, or adverbial clause [if any] – one is
creating. (Maschler 2018: p. 698)
Elaborative insubordinate she-clauses are formatted with both appeal and non-
appeal intonation contours. We begin with the more common variety.
4.1 Elaborative insubordinate she-clauses in non-appeal intonation contours
The great majority of elaborative insubordinate she-clauses in non-appeal intona-

tion contours (79 out of 86, 92%) are produced turn-medially. They may elaborate
or specify a prior noun phrase or an entire clause, and the data show that insubor-
dinate she-clauses may also elaborate gestures.
14. See Lehti-Eklund, 2002; Anward, 2004; Verstraete, D’Hertefelt & Van Linden, 2012; Weinert,
2012; Mertzlufft & Wide, 2013; Günthner, 2014; Wide, 2014; Cristofaro, 2016.
100 Yael Maschler
4.1.1 Elaborating an NP
In the following excerpt, from a different part of the ‘Chinese Poetry’ interaction,
Dotan, who works as a tour guide, tells Alex how the Australian tourists in his
recent group thanked him at the end of their three-week tour:
(3) ‘Australians’
123 DOT: ...ha-xamudim ha-'ele 'asu l-i--,
DEF-cuties DEF-this.PL do.PST.3PL to-1SG
...those cuties made me--,
124 ...plakat,
poster
...a poster,
125 ...'im-- 'e--m--,
...with-- u--hm--,
{dim., creaky voice}
126 she-katvu l-i toda raba. x
that-write.PST.3PL to-1SG thank_you very_much
that they wrote me “thank you very much.”
dot:
dot: xsmiles
smiles
127 *(1.69)*
ale: *shakes head and pulls it up in bewildered expression*
Alex Dotan
fig. 7
fig. 8
128 DOT: .....hem
.....they
129 xke'ilu kmo-- m--,
as_if like m
like a kind of m--,
dot: xdemonstrates twice opening a book with both hands--->
130 sefer-maxzor ka-ze?,x*

yearbook like-this
yearbook sort of?,
dot: -->---------------------x
ale: *nods, smiles
fig. 9
like a kind of m--,
dot: xdemonstrates twice opening a book with both hands--->
130 sefer-maxzor ka-ze?,x*

yearbook like-this
yearbook sort of?, Chapter 4. The insubordinate – subordinate continuum 101
dot: -->---------------------x
ale: *nods, smiles
fig.
fig. 9 9
131 xshe-kulam kotvim 'e--,

that-everybody write.PRS.M.PL uhm
that everybody writes u--hm,
{creaky voice}
dot: xdemonstrates writing with left hand-->
fig. 10
132 braxa?, x
blessing
a blessing?,
dot: -->------ x
133 ...la-yom huledet?,

to.DEF-day birth
...for birthdays?,
134 ALE: *'okey.
okay.
ale: *nods
135 DOT: x(1.4) x
dot: xinbreathx
136 ...'az hem 'asu l-i ka-ze?,

do they do.PST.3PL to-1SG like-this
so they made me one of those?,
Dotan (sitting on the right (Figures 7–10)) tells Alex that the tourists in his group,
whom he refers to as ‘those cuties’, had made a poster for him at the end of the tour
in order to thank him (3. 123–126).15 In response to Dotan’s telling, Alex pulls
back while moving his head from side to side, in a gesture conveying mild bewil-
derment (3. 127, Figures 7–8). Dotan responds to this with a further elaboration
of his previous utterance following the discourse marker ke'ilu ‘like’ (which often
precedes elaborations (Maschler, 2002)): ke'ilu kmo-- m--, sefer maxzor kaze?, ‘like
a kind of m--, year book sort of?,’ (3. 129–30) with continuing appeal intonation.
This type of intonation contour ends in a rise in pitch, not quite as high as that
15. The insubordinate she-clause at line 126 is discussed in excerpt (9) below.
102 Yael Maschler
of appeal intonation, followed by a slight fall in pitch (see Figure 11), and signi-
fies (see footnote 12) the speaker’s intention to continue while at the same time
appealing for some minimal response from the recipient (Du Bois, forthcoming,
§5.4.3: p. 3).
0.1138
-0.1057
0 0.7979
Time (s)
300
250
200
150
Pitch (HZ)
100
50
se fer max zor ka ze
year book sort of
0 0.7979
Time (s)
Figure 11. ‘Australians’, line 130.
In response to this intonation contour, Alex nods and smiles (3. 129, Figure 9).
Note that no pause separates lines 130 and 131. Line 130 is immediately followed by
Dotan’s she-clause she-kulam kotvim 'e--h, braxa?, la-yom huledet?, ‘that everyone
writes u--h, a blessing?, for birthdays’?, accompanied by his gesture demonstrating
writing with his left hand (3. 131–133, Figure 10). Dotan is conflating two Israeli
customs, both of which involve writing a special note for a friend: (1) the yearbook
custom – writing notes in a book published at the end of high school, and (2) the
birthday card custom – writing a note (referred to as braxa ‘a blessing’) on a card
to accompany the gift. In any event, in response to these two additional continuing
appeal intonation contours (3. 132–133) Alex nods his head in recognition of the
custom, whichever of the two it may be, and responds with okay (3. 134), and,
following an inbreath and pause (3. 135–136) Dotan continues the telling.
The she-clause she-kulam kotvim 'e--h, braxa?, la-yom huledet?, ‘that everyone
writes u--h, a blessing?, for birthdays’? thus elaborates the hedged NP sefer maxzor
kaze ‘a yearbook sort of ’16 by specifying its use. It is classified here as insubordi-
nate in the sense of being a loosely-integrated she-clause, because it lacks some
of the morphosyntactic dependency marking – i.e., the normative resumptive
pronoun (e.g., ‘a yearbook sort of [in which] everyone writes …’) – characterizing
relative clauses.
Of the 98 elaborative insubordinate she-clauses in the database, 31 (32%)
elaborate a noun phrase. Only 4 (13%) are produced as increments, while the
great majority (N=25) follow a continuing intonation contour. Of those 25, only 5
follow a continuing appeal intonation contour, as in Excerpt (3).17
4.1.2 Elaborating a clause

An additional 32 insubordinate she-clauses (33%) elaborate a clause, as in
Excerpt (4), in which Nurit is recounting to three friends the content of a book
entitled ‘Maiden Farm’, a sequel to a book they have all read:
(4) ‘Maiden Farm’
103 ORI: rega,
[just a] moment,

104 lo hevanti ma kore 'im sasha
not understand.past.1sg what happen.prs.sg.m with Sasha
I didn’t understand what happens with Sasha

105 ..'im fanya.
with Fania.

106 NUR: ...s.. ha--sefer ke'ilu,
b def-book like
b the-- book like,

107 be-xavat ha-'almot,
in-farm.poss def-maidens
in Maiden Farm,

108 mamshi--x?,
continue.prs.sg.m
continue--s?,

109 ...hi--
...she--

16. For the hedging function of kaze lit. ‘like this’, see Maschler (2001).
17. This is thus different both syntactically and prosodically from the cases discussed in
Stoenica & Pekarek Doehler, this volume, in which a full-fledged relative clause (missing no
morphosyntactic dependency markers) is produced always following ‘try-marked’ intonation
(Sacks & Schegloff, 1979) (termed here ‘continuing appeal intonation’).
104 Yael Maschler
➔110 ..she-fanya kvar 'e

that-Fania already uh
that Fania is already uh

111 ..b 'em
b uhm

112 ...mevuge--ret,
mature.sg.f
matu--re,

113 yoledet 'od yeladi--m,
give_birth.prs.sg.f more childre--n,
gives birth to more childre--n,
(continues to line 145)
In response to Orit’s question, following a short false start hi-- ‘she--’ (4. 109),
Nurit self-repairs beginning with she-. The she-clause complex she-fanya kvar 'e, b
'em mevuge--ret, yoledet 'od yeladi--m, ‘that Fania is already uh, b u--hm matu--re,
gives birth to more childre--n,’ (4. 110–113), as well as the following 22 intonation
units (not shown), elaborate the clause ha--sefer ke'ilu, be-xavat ha-'almot, mamshi-
-x?, ‘the-- book like, in Maiden Farm, continues?,’ (4. 106–108) by specifying the
manner in which ha-sefer mamshix ‘the book continues’ in the sequel; however,
this rather long clause complex also specifies the content of the continuation itself.
It is problematic to consider this clause complex an object complement of mam-
shix ‘continues, goes on’, because the verb is intransitive in this meaning. However,
it is also problematic to consider it an adverbial clause, because of the lack of a
manner conjunction, such as be'ofen she- ‘in a manner that’ or kax she- ‘such that’.
It is classified here as a loosely-integrated she-clause, because it does not fall in
any of the categories – relative, object or adverbial complement. Furthermore, the
elaboration continues for another 22 intonation units, raising serious doubts as to
the status of the clause ha-sefer mamshix ‘the book continues’ as a matrix clause.
Other loosely-integrated she-clauses elaborating clauses and ‘missing’ adver-
bial conjunctions in the database are illustrated in Excerpts (5)–(7):
In Excerpt (5) Daria recounts a story she had heard from Inbal, a female co-
worker. In the following part of the story, Daria reports how Inbal begins to reveal
to Eli, a man who had been ‘planted’ at their office in order to investigate drug
usage, that she knows he was sent to spy on them:
(5) ‘Collaborator’
186 DAR: hi 'osa l-o,
she do.prs.sg.f to-3sg.m
she says to him,

187 'al taxshov she-'ani metumtemet,
neg think.fut.2sg.m that-I dumb
don’t think that I’m dumb,

188 ..'ani yoda'at!

I know.prs.sg.f
I know!

189 'ani yoda'at ma 'ata 'ose,
I know.prs.sg.f what you.2sg.m do.prs.sg.m
I know what you’re doing,

190 ma 'ata xoshev 'ani mefageret,
what you think.prs.sg.m I retarded.sg.f
what do you think I’m an idiot,

➔ 191 she-'ani lo yoda'at she-'ata mashtap?
that-I neg know.prs.sg.f that-you collaborator
that I don’t know you’re a collaborator?
The she-clause she-'ani lo yoda'at she-'ata mashtap ‘that I don’t know you’re a
collaborator’ (5. 191) elaborates the clause 'ani mefageret ‘I’m an idiot’ (5. 190)
by spelling out the extent of the ‘dumbness’: what do you think I’m dumb, [to
the extent that] I don’t know you’re a collaborator? This cannot be considered an
adverbial complement of the nominal clause I’m an idiot, because of the lack of
an adverbial conjunction (such as 'ad kedey kax she- ‘to the extent that’). The she-
clause elaborates the preceding clause by specifying the extent of the predication
made by it – in what sense she could be ‘an idiot’.18
Excerpt (6) illustrates an elaboration of time:
(6) ‘Presents from Abroad’
86 OSN:
...ze mi-pa'a--m,
this from-once
this is from a while ba--ck,

➔ 87 ...she--,
tha--t,

88 ..la-'anashim ba-'arets lo haya kesef.
to.def-people in.def-country neg be.pst.3sg.m money
people here didn’t have any money.
The she-clause she--la-'anashim ba-'arets lo haya kesef ‘that the people here didn’t
have any money’ (6. 87–88) elaborates the preceding clause by expanding on the
time at which the predication ze mi-pa'am ‘this is from a while back’ (6. 88) took
place. Again, it cannot be considered an adverbial clause, because the appropriate
adverbial conjunction (e.g., kshe- ‘when’) is ‘missing’.
In Excerpt (7) we find another, slightly different example:
18. The syntactic analysis of line 191 as being an alternative complement (or list item) to ma
'ata xoshev ‘what do you think’ is ruled out on the basis of prosody – the fact that 'ani mefageret
‘I’m an idiot’ is part of the same intonation unit as the matrix clause. Furthermore, had it been
the first member of a list, we would expect it to also be preceded by a she- complementizer, thus
manifesting syntactic ‘balance’ between the two list members.
106 Yael Maschler
(7) ‘Siamese Twins’

201 MIC: @bo'u @titstalmu [tstalmu] @'it-i.
come.impt.2pl be_photographed.fut.2pl with-1sg
come have your picture taken with me.

➔ 202 ...@she-yera'e @ke-'ilu @yesh @l-i @xaverim.
that-seem.fut.3sg.m as-if there_is to-1sg friend.pl
[so] that it seems as if I have friends.
The she-clause she-yera'e ke'ilu yesh li xaverim ‘that it seems as if I have friends’
elaborates the clause bo'u titstalmu 'iti ‘come have your picture taken with me’
by spelling out the purpose for this request. As in the previous four excerpts, it
cannot be considered an adverbial clause, because the appropriate adverbial
conjunction – this time kedey she- ‘so that’ – is ‘missing’. Here the insubordinate
she-clause is produced as an increment, following sentence final falling intonation
(7. 201, Figure 12).
1
1
0
0
-1
-1 0 0.9765
0 Time (s) 0.9765
Time (s)
500
500
400
400
300
300
) Z)
(HZ(H
200
Pitch
200
Pitch
75
75 bo ’u tstal mu ’i ti
bo ’u tstal mu ’i ti
come have your picture taken with me
come have your picture taken with me
0 0.9765
0 Time (s) 0.9765
Figure 12. ‘Siamese Twins’ line 201 Time (s)
Of the 32 tokens elaborating clauses, only 7 (22%) are produced as increments.
4.1.3 Elaborating a gesture

Following the evaluation she-ze dey pashut, ‘which is pretty simple’ of excerpt (1),
Alex proceeds to illustrate the simplicity of the method of transliterating Chinese
tones in romanization in a four-part list. Each member of this list is a depiction of
the transliteration of a different tone:
(8) ‘Chinese Poetry’(Excerpt (1) continued)

122 ALE: she-ze dey pashut.
that-this.M.SG pretty simple
which is pretty simple.
123 *ke'ilu kav... ka-ze* ze--.. *flet*?,

as_if line like-this.M.SG this.M.SG flat
like a line like this i--s flat?
{creaky}
ale: *straight line with left index finger* *straight line
with left hand*
124 ...*ma--*,
...ma--,
{creaky voice}
ale: *straight line with left hand*
Alex Dotan
fig. 12
125 ...'e--m,
...u--hm,
{creaky voice, p}
126 ...*'im ze.. ka-ze* chupchick *'ole--*,

if this.M.SG like-this.M.SG little_stroke go_up.PRS.M.SG
...if it’s like a little stroke going up,
{----------creaky voice, p----------} {-f-}
ale: *diagonal line going up with left index finger* *diagonal
line going up with left index finger*
fig. 13
127 'az
then
128 ...*/'a/*
{pp}
ale: *diagonal line going up with index finger*
129 */?/'ata 'ole ba-kol?,*

you.M.SG go_up.PRS.M.SG in.DEF-voice
/?/you go up with [your] voice?,
ale: *open left palm moving from near his throat up and out*
108 Yael Maschler
ale: *open left palm moving from near his throat up and out*
fig. 14
130 *'im /ze/ chupchik yored,*

if /this.M.SG/ little_stroke go_down.PRS.M.SG
if /it’s/ a little stroke going down,
{---------------acc--------------}
ale: *diagonal line going down with left index finger*
131 'ata yored,

you.M.SG go_down.PRS.M.SG
you go down,
{-----acc-----}
132 ve-yesh la-hem *ka-ze,*

and-there_is to-3M.PL like-this.M.SG
and they have [one] like this,
ale: *left-to-right semi-circle with right index finger*
fig. 15
133 *(0.54)*
ale: *inbreath,right-to-left semi-circle with right index finger*
fig. 16
fig. 16
fig. 17
134 *she-yore
that-go_dow[n].PRS.M.SG
that goes dow[n]
ale: *left-to-right semi-circle with right index-->
fig. 18
135 yored* *ve-'ole.*

go_down.PRS.M.SG and-go_up.PRS.M.SG
goes down and goes up.
ale: -->finger* *top-down vertical line with right index finger*
136 ...*ve-zehu.*
and-that’s_it
...and that’s it.
{staccato}
ale: *back to home position*
137 DOT: ...zehu?

that’s_it
...that’s it?
{staccato}
138 ALE: ...ken.

...yes.
{staccato}
139 DOT: 'ah.

oh.
{high pitch, staccato}
140 ...'okey.
...okay.
The clause she-yored ve-'ole ‘that goes down and goes up’ (8. 134–135) is not a
canonical, syntactically integrated relative clause (although it is the closest to the
syntactically integrated end of the continuum of all tokens examined so far): this
clause is somewhat separated from the preceding intonation unit ve-yesh la-hem
ka-ze lit. ‘and they have like this’ (8. 132) by Alex’s inbreath (8. 133), but more
crucially, the preceding clause (8. 132) contains no NP antecedent.19
The she-clause follows Alex’s gesture which he began while articulating ve-yesh
la-hem ka-ze lit. ‘and they have like this’ (8. 132). Alex switches at the beginning
of intonation unit 132 from gesturing with his left hand (as he had done for the
first three types of Chinese tones, Figures 12–14) to gesturing with his right hand
(Figure 17) for depicting the form of the fourth diacritic on his list. This gesture
is performed by drawing a semi-circle in the air, going down and then up, from
left to right, while Alex says ka-ze ‘like this’20 (8. 132, Figure 15), and then making
the same gesture in the opposite direction, from right to left, during his following
inbreath (8. 133, Figure 16). The third iteration of this gesture – this time from left
to right – begins at the start of the she-clause she-yored ve-'ole ‘that goes down and
goes up’ (8. 134, Figure 18).
19. While it is true that in traditional grammatical terms the deictic kaze (composed of the
preposition ke- ‘as’ and the demonstrative ze ‘this’, ‘like this’) can be understood to ‘imply’ a
noun such as chupchik ‘little stroke’, this noun was uttered two clauses earlier, rendering this a
loosely-integrated relative clause.
20. Cf. English ‘like this’ in Streeck’s study: “like buys the speaker the option of continuing the
utterance with an enactment of some sort, to switch from the verbal to the nonverbal mode”
(Streeck, 2002: pp. 585–586).
110 Yael Maschler
This gesture is iconic of the shape of the diacritic, but the actual diacritic is, of
course, a static entity. The verbs yored ‘goes down’ and 'ole ‘goes up’ therefore refer
to the motion of the line while this diacritic is being produced, or even to the act
of writing it, both of which the speaker embodies via his gestures. This she-clause
can therefore be seen to elaborate the previous stretch of embodied discourse by
spelling out in words the actual contours of the gesture: she-yored ve-'ole ‘that goes
down and goes up’.
Alex’s self-repair at she-yore yored ‘goes dow[n] goes down’ (8. 134–135) al-
lows perfect synchronization with his semi-circle, so that the semi-circle gesture
is completed precisely at the end of yored ‘goes down’ (8. 135). He withdraws from
his action by gesturing a vertical line while uttering ve-'ole ‘and goes up’ (8: 135)
and completing his she-clause, followed by the discourse marker ve-zehu ‘and
that’s it’ (8: 136); back in home position (Sacks & Schegloff, 2002), he marks the
end of the four-part list. Thus we see that Alex’s insubordinate she-clause emerges
“incrementally and responsively to interactional contingencies, among them
embodied ones” (Keevallik 2018: 17).
While the previous excerpt manifested a deictic element tying the gesture to
the actual diacritic in the extralingual world, a she-clause may also elaborate a
gesture unaccompanied by any talk. This is found in ‘Australians’ (Excerpt (3), the
beginning of which is reproduced below as Excerpt (9)):
(9) ‘Australians’
123 DOT: ...ha-xamudim ha-'ele x'asu l-i--,x
DEF-cuties DEF-this.PL do.PST.3PL to-1SG
...those cuties made me--,
dot: xmarks an area on the table with both hands, looks downx
Alex Dotan
fig. 19
fig. 19
124 ... xplakat,x

poster
...a poster,
dot: xmarks an area on the table with both hands, looks at Alexx
fig. 20
fig. 19
124 ... xplakat,x
poster
...a poster, Chapter 4. The insubordinate – subordinate continuum 111
dot: xmarks an area on the table with both hands, looks at Alexx
fig.fig.
20 20
125 ... x'im-- 'e--m--,

... with-- u--hm--,
{dim., creaky voice}
dot: xdemonstrates writing with left hand-->
fig. 21
126 she-katvu l-i toda raba.x x

that-write.PST.3PL to-1SG thank_you very_much
that they wrote me “thank you very much.”
dot: -->----------------------------------------
dot: -->----------------------------------------smiles-->
x xsmiles-->
127 *(1.69)*
ale: *shakes head and pulls it up in bewildered expression*
dot: -->-------------------------------------------------x
fig. 22
Dotan’s telling about the poster he was given at the end of the tour ha-xamudim
ha-'ele, 'asu li plakat, 'im-- 'e--m ‘those cuties, made me a poster, with-- uh--m’ (9.
123–125) is immediately followed by the she-clause she-katvu li toda raba ‘that
they wrote me “thank you very much”’ (9. 126). This clause is loosely integrated
with the preceding clause only via the she- component. It cannot be considered a
relative clause modifying plakat ‘poster’, because it is separated from plakat by the
preposition 'im ‘with’, and prepositions do not precede relative clauses in Hebrew.
Furthermore, no resumptive pronoun tying back to ‘poster’ is to be found. This
she-clause cannot be considered an object or adverbial complement either, because
112 Yael Maschler
it cannot be interpreted as modifying the clause ha-xamudim ha-'ele, 'asu li plakat

‘those cuties, made me a poster’ (9. 123–124).
Note that line 125 'im-- 'e--m ‘with-- uh--m--’ is not a fragmentary intonation
unit but rather a continuing one, and that its creakiness continues all the way to
the beginning of the following intonation unit (line 126 begins with no pause).
Thus, no self-repair is involved here. Examination of the embodied behavior
reveals that these two intonation units cohere not only through prosody but also
by means of gesture: while Dotan verbalizes plakat ‘poster’ (9. 124), he makes a
gesture depicting the borders of a poster on the desk in front of him (Figure 20).
Line 125 is then uttered in drawled out fashion while Dotan gestures the act of
writing (Figure 21), a gesture he continues well past the end of his verbalization
of 'im-- 'e--m ‘with-- uh--m--’, all the way to the end of line 126. The immediately
following she-clause therefore elaborates Dotan’s writing gesture by explicating the
gesture: 'asu li plakat, 'im-- 'e--m ‘they made me a poster, with uh--m [writing
gesture]’ ‘and this gesture I just started making means ‘that they wrote me “thank
you very much” ’. Thus, insubordinate she-clauses can elaborate not only NPs and
clauses, but also gestures.
4.2 Elaborative she-clauses in appeal intonation
Of the 98 elaborative she-clauses throughout the data, 12 (12%) are produced with
an appeal intonation contour. All of them occur turn-initially. A third (N = 4)
request elaboration of a prior utterance of the interlocutor’s, whereas the majority
(N = 8) request confirmation of an elaboration offered by the speaker concerning
a prior utterance of the interlocutor’s.
4.2.1 Requesting elaboration

In Excerpt (10) Oleg begins to tell his two friends Ella and Yuri what he did that
day in order to start receiving his unemployment money from Social Security:
(10) ’Unemployment Money’
2 OLE:
...sof sof 'asiti,
end end do.past.1sg
I finally did,

3 'et kol ha-dvarim,
acc all def-things
all the things,

4 she-hayiti tsarix la'asot hayom.
that-be.past.1sg have_to do.inf today
that I was supposed to do today.

5 ELL: ....ma 'asita?

what do.past.2m.sg
what did you do?

6 OLE: 'et kol ha-dvarim,
acc all def-things
all the things,

7 she-hayiti tsarix la'asot hayom.

8 ELL: ...'okey.
okay.

9 OLE: ...mizman.
a long time ago.

➔ 10 YUR: ...she-ze?
that-this.m.sg
which is?

11 OLE: halaxti--,
go.past.1sg
I we--nt,

12 le--,
to--,

13 ..halaxti le-bituax le'umi?,
go.past.1sg to-insurance national
I went to [the] Social Security [office]?,

14 ...sof sof?,
end end
finally?,

15 ...ve-haya sagur.
and-be.past.3m.sg closed.m.sg
and it was closed.
(continues narrative)
In response to Oleg’s announcement that he finally did all the things he was sup-
posed to do that day (10. 2–4), Ella seeks clarification ma 'asita? ‘what did you do?’
(10. 5), and he repeats verbatim 'et kol ha-dvarim, she-hayiti txarix la'asot hayom.
‘all the things, that I was supposed to do today’ (10. 6–7) in sentence-final falling
intonation (10. 7). Following Ella’s receipt token 'okey (10. 8), he then adds an
increment self-repairing hayom ‘today’ to mizman ‘a long time ago’ (10. 9). At this
point Yuri asks she-ze? ‘which is?’ (10. 10) in appeal intonation (Figure 23):
This is a request for elaboration of the NP kol ha-dvarim, she-hayiti txarix
la'asot ‘all the things, that I was supposed to do’ (10. 6–7). Indeed, Oleg responds
to this request for elaboration by naming one of the things he was supposed to do:
halaxti--, le--, halaxti le-bituax le'umi?, sof sof?, ‘I we--nt, to--, I went to [the] Social
Security [office]?, finally?,’ followed by a narrative of what happened to him there.
114 Yael Maschler
0.2355
-0.2372
0.37 0.75
Time (s)
250
200
150
Pitch (HZ)
100
50
she ze
which is
0.37 0.75
Time (s)
Figure 23. ‘Unemployment Money’, line 10
Yuri’s stand-alone she-ze? ‘which is?’ (10. 10), composed of she- followed by
the masculine singular demonstrative pronoun ze (which can function as a copula
in Hebrew, cf. 2. 29), is syntactically unintegrated into any matrix clause. Produced
with appeal intonation here, it solicits an elaboration of Oleg’s immediately
preceding NP.
4.2.1 Requesting confirmation of candidate elaborations

Turn-initial insubordinate she-clauses with appeal intonation may also be employed
to request confirmation of candidate elaborating completions of an interlocutor’s
immediately preceding utterance. An example of this appears in Excerpt 11, in
which Gadi begins to explain why his interest in a certain woman is not met with
mutual interest on her part:
(11) ‘Too Young’
75 GAD: ...'exad ha-dvarim,
one DEF-things
one of the things,
76 she-haxi kaze--,
that-most like
that most li--ke,
77 /???/
78 ORE: ma
75 GAD: ...'exad ha-dvarim,
one DEF-things Chapter 4. The insubordinate – subordinate continuum 115
one of the things,
76 she-haxi kaze--,
that-most like
that most li--ke,
77 /???/
78 ORE: ma
pu[t]
79 martia 'ota?
put_off.PRS.SG.M ACC.3SG.F
puts her off?
80 GAD: ..ken.
yes.
81 RON: ...she-ma?
that-what
that what?
82 ORE: she-'ata katan yoter?

that-you young.SG.M more
that you are younger?
83 RON: ...she-'ata katan?

that-you young.SG.M
that you are young?
84 GAD: /ken/.
/yes/.
Gadi begins this delicate matter with the drawled out 'exad ha-dvarim, she-haxi
kaze--, ‘one of the things, that most li--ke,’ (11. 75–76). Oren co-constructs the
remainder of Gadi’s utterance with martia 'ota? ‘puts her off?’ (11. 78–79) in ap-
peal intonation, proffering a continuation of this relative clause.21 Upon Gadi’s
confirmation (11. 80), Rona employs the insubordinate clause she-ma? ‘that what?’
(11. 81) in appeal intonation, soliciting an elaboration (cf. Excerpt 10) of what it
is that ‘puts the woman off ’ (11. 75–79). Oren overlaps Rona with the candidate
elaboration of Gadi’s utterance she-'ata katan yoter? ‘that you are younger?’ (11.
82) with appeal intonation, and Rona follows with an almost identical candidate
(11. 83), in answer to her own question of 11. 81. Gadi responds to both of these
candidate elaborations affirmatively (11. 84). The overlap renders Gadi’s response
only barely audible, but the continuation of the talk (not shown) confirms
this interpretation.
The she-clauses at (11. 82, 83) are syntactically unintegrated into any matrix
clause and are therefore considered here insubordinate. Produced with appeal
intonation, they each inquire about the adequacy of the candidate elaboration.
21. This is a full-fledged Hebrew relative clause, with the resumptive pronoun in the form of the
subject pronoun attached to the verb martia ‘puts off ’ and the antecedent 'exad ha-dvarim ‘one
of the things’.
116 Yael Maschler
5. Discussion and conclusion
This chapter has examined a variety of Hebrew she-clauses which do not fit
traditional accounts of subordination, either because they are syntactically un-
integrated (unembedded in any matrix clause) (Excerpt 1, 2, 10, 11) or because
they are syntactically loosely integrated (i.e., cannot be viewed unambiguously as
a relative, complement, or adverbial clause because some morphosyntactic depen-
dency marker is ‘missing’ (Excerpt 3, 4, 5, 6, 7, 8, 9)).
The insubordinate she-clauses examined here vary in terms of their prosodic
features and contexts. Only 26 out of the 113 evaluative and elaborative insubordi-
nate she-clauses throughout the data (23%) are produced as increments. The great
majority (N = 68, 60%) are produced turn-medially following usually continuing,
but sometimes continuing appeal intonation and therefore do not constitute
increments. However, there is a difference here between evaluative insubordinate
she-clauses and elaborative ones. For the evaluative variety, 8 out of 15 (53%) are
produced as increments, whereas for the elaborative ones, this figure is only 21%
(18 of 86 tokens produced in non-appeal intonation). The action of evaluating a
previous stretch of talk, then, is more often added on following prosodic comple-
tion and is thus more markedly separated from the action carried out immediately
preceding it, while the action of elaborating prior talk is far more often produced
following a continuing intonation contour, in a manner prosodically continuous
with the previous action. This is not surprising, considering the fact that when
elaborating a previous stretch of talk, the speaker remains within the referential
realm, whereas in evaluating a previous stretch of discourse there is always a move
from the referential to the interpersonal realm of discourse (involving relations
of speaker to text (Becker, 1979)). This more prominent shift is thus iconically
reflected in the more salient prosodic boundary.
Only 12 insubordinate she-clauses in our corpus (12%) are produced with
appeal intonation. In those cases they are always turn-initial and request (or offer
a candidate) elaboration of the interlocutor’s immediately prior talk (Excerpt 10,
11). These clauses, as well as the evaluative cases, are always syntactically unin-
tegrated, whereas in the elaborative category there is a greater tendency towards
loose integration, reflected in the greater prosodic integration of the latter.
All insubordinate she-clauses produced with non-appeal intonation (N = 101)
share a meta-level action: she- is employed as a ‘wildcard’ tying back to an im-
mediately prior stretch of interaction and projecting an elaboration or evaluation of
it, without much concern about which particular type of complex construction is
being created, if any. In the case of elaboration, the elaborated stretch of discourse
preceding the she-clause may be an NP (Excerpt 3), a clause (Excerpt 4–7), or a
gesture performed by the speaker (Excerpt 8, 9). Speakers’ minute multimodal
synchronizations of verbal and embodied conduct accompanying the production

of their insubordinate she-clauses, as seen in these latter two examples, strongly
support a reconceptualization of grammar “as an assembly of routinized methods
for the organization of vocal conduct, capable of incorporating aspects of par-
ticipants’ bodily behavior” (Keevallik, 2018: p. 1, and see also, e.g., Kendon, 2004;
Streeck, 2009, 2013; Mondada, 2016; Kärkkäinen and Thompson 2018).
We have seen that by far the great majority of insubordinate she-clauses are
elaborative. However, upon examination of a canonical relative clause such as
(10. 4), it quickly becomes evident that integrated she-relative clauses in bi-clausal
constructions also elaborate immediately preceding discourse:
(10) ’Unemployment Money’(partial)
2 OLE:
... sof sof 'asiti,
end end do.past.1sg
I finally did,

3 'et kol ha-dvarim,
acc all def-things
all the things,

➔ 4 she-hayiti tsarix la'asot hayom.
The relative clause at line 4 elaborates the head kol ha-dvarim ‘all the things’, speci-
fying that those ‘things’ are things the speaker ‘was supposed to do that day’.22 In
fact, any attribute specifies the head it modifies. Similarly, the canonical comple-
ment clause at (5. 187) is also an elaboration of immediately preceding discourse:
(5) ‘Collaborator’ (partial)
186 DAR: hi 'osa l-o,
she do.prs.sg.f to-3sg.m
she says to him,

➔ 187 'al taxshov she-'ani metumtemet,
neg think.fut.2sg.m that-I dumb
don’t think that I’m dumb,
The canonical complement clause 'ani metumtemet ‘I’m dumb’ elaborates the
negated matrix verb 'al taxshov ‘don’t think’ by naming what the interlocutor is
not supposed to think – just as any argument or adverbial complement elaborates
22. There is no resumptive pronoun here, but this is considered a canonical relative clause
according to traditional Hebrew grammar, because an accusative resumptive pronoun need
not occur. For an elaborative full-fledged relative clause including a resumptive pronoun, see
Excerpt (11. 75–79) and footnote 21.
118 Yael Maschler
the action/event described by the verb by spelling out the referents involved in the
action/event or the circumstances in which it takes place.
In previous approaches insubordinate clauses have been viewed as imperfect
realizations of the canonical ‘subordinate’ variety resulting from the disintegration
of the patterns of complex syntax (Cristofaro, 2016: p. 419, see Section 1 above).
The present study supports the opposite view. Since canonical, syntactically
integrated she-clauses perform the same actions – elaboration (and sometimes
evaluation) – accomplished by the unintegrated or loosely integrated variety,
canonical ‘subordinate’ clauses may be regarded as a sub-type of the latter. The
data suggest viewing canonical, syntactically integrated ‘subordinate’ she-clauses
as grammaticizations in which elaborative (and to some extent evaluative) actions
have crystallized to form the bi-clausal relative, complement, and adverbial clause-
combining patterns familiar to us from traditional complex syntax23. Further
evidence supporting this suggestion comes from the fact that not only in Hebrew
but in many of the world’s languages (Hendery, 2012: pp. 97–117), we find the
same morpheme at the basis of the three relevant types of conjunction – relativizer,
complementizer, and various adverbial conjunctions. This can be explained by the
finding that all of these cases can be seen to perform the same function of tying
back to an immediately prior stretch of talk or gesture and projecting – via she- or
its equivalents in other languages – its continuation by elaboration or evaluation.24
Another piece of evidence supporting the grammaticization of the integrated
variety from the unintegrated one is the fact that in many languages, as in Hebrew
(see Section 1), the relativizer originates from a demonstrative pronoun. It is only
natural for a pointing action to then be followed by an elaboration of the entity
pointed at. In line with principles of grammaticization (e.g., Bybee, 2003; Hopper
& Traugott, 2003), the two co-occurring actions of pointing and elaborating may
over time lead to the grammaticization of a demonstrative into a relativizer.
All this leads to a scalar conception of she-clauses, according to which they
may be positioned along a continuum of integratedness stretching from insubor-
23. A related point was made for English and German bi-clausal constructions of another
variety – pseudo-clefts (Hopper and Thompson, 2008).
24. Of course once canonical syntactically integrated constructions have crystallized in the
language, they may continue to undergo linguistic change. For example, once adverbial con-
junctions have crystallized from specific NPs followed by a she-clause (see Maschler, 2011), they
may get reduced back to she- (as in kedey she- ‘so that’ > she-, or kshe ‘when’ > she-) as part of a
cyclic process. However, reduction cannot be the whole story, as it is incapable of providing an
account for the fact that the same morpheme – she- – is found at the basis of the three relevant
types of conjunction in the first place. For another argument against the exclusivity of the reduc-
tion explanation, see Maschler (2018: pp. 685–787).
dinate to canonical ‘subordinate’ tokens. The various cases termed ‘insubordinate’

in the present study vary in their placement along this continuum (Figure 24):
Syntactically Syntactically Syntactically Syntactically Syntactically Syntactically Syntactically

unintegrated unintegrated loosely loosely loosely loosely integrated integrated
she-clause in she-clause, integrated she- integrated integrated she-clause, canonical
appeal unembedded clause, she-multi- she-clause, + resumptive Pro, ‘subordinate’
intonation in any - NP clause - adverbial - NP antecedent she-clause
matrix antecedent, construction, conjunction (relative,
clause - resumptive - adverbial complement, or
(‘cosubordinate’) Pro conjunction adverbial)
Syntactically
loosely integrated
she-clause,
- resumptive
Pro
Figure 24. The insubordinate – subordinate continuum of she-clauses
Syntactically unintegrated she-clauses verbalized with appeal intonation

(Excerpt 10, 11) are positioned at the end (see Figure 24) opposite that of canoni-
cal syntactically integrated cases (Excerpt (10. 4), (11. 75–79), (5. 187)). In fact,
one could question whether these still count as cases of clause-combining, or
whether perhaps the she- in this variety should rather be considered a discourse
marker. Immediately to the right of these syntactically unintegrated cases would
be the syntactically unintegrated she-clauses which are unembedded in any matrix
clause (‘cosubordinate’ (Van Valin, 1984), Excerpt 1, 2). Next come the syntacti-
cally loosely integrated tokens exhibiting no NP antecedent and no resumptive
pronoun (Excerpt 9). Further yet to the right would be syntactically loosely inte-
grated she-multi-clause constructions with no adverbial conjunction (Excerpt 4),
and then syntactically loosely integrated she-clauses in bi-clausal constructions
exhibiting no adverbial conjunction (Excerpt 5–7) or no resumptive pronoun
(Excerpt 3). Loosely integrated clauses with a resumptive pronoun but no NP
antecedent would come next (Excerpt 8), closest to the syntactically integrated
canonical ‘subordinate’ variety. The further along we move towards the right end
of this continuum, the fuller is the morphosyntactic dependency marking. The
basic meta-level actions accomplished by all these types of clauses, however, re-
main constant. Movement towards the right on this continuum thus represents the
“continual movement towards structure” (Hopper, 1987: p. 142) characterizing
the emergence of Hebrew complex syntax from interaction.
120 Yael Maschler
References
Anward, Jan. (2004). ‘att’ [‘that’]. Språk och stil 13, 65–85.
Auer, Peter. (2005). Projection in interaction and projection in grammar. Text 25, 7–36.
Auer, Peter. (2007). Why are increments such elusive objects? An afterthought. Pragmatics
17(4), 647–658. https://doi.org/10.1075/prag.17.4.03aue
Auer, Peter. (2009). On-line syntax: Thoughts on the temporality of spoken language. Language
Sciences 31,1–13. https://doi.org/10.1016/j.langsci.2007.10.004
Becker, Alton L. (1979). Text-building, epistemology, and esthetics in Javanese shadow theater.
In Alton L. Becker & Aram Yengoyan (Eds.), The imagination of reality (pp. 211–243).
Norwood, N. J. : Ablex.
Bergsträsser, G. (1909). Das hebräische Präfix še,” ZAW, 29, 40–56.
Blau, Yehoshua. (1966). yesodot hataxbir [Foundations of syntax]. Jerusalem: hamaxon ha'ivri
lehaskala bixtav.
Boersma, Paul & Weenink, David. (2019). Praat: doing phonetics by computer [Computer pro-
gram]. Version 6.1, retrieved 13 July 2019 from http://www.praat.org/
Bybee, Joan. (2003). Mechanisms of change in grammaticization: The role of frequency. In Brian
D. Joseph and Richard D. Janda (Eds.), The handbook of historical linguistics (pp. 602–623).
Oxford: Blackwell. https://doi.org/10.1002/9780470756393.ch19
Chafe, Wallace. (1994). Discourse, consciousness, and time: The flow and displacement of con-
scious experience in speaking and writing. Chicago: University of Chicago Press.
Clift, Rebecca. (2007). Grammar in time: the non-restrictive ‘which’-clause as an interactional
resource. Essex Research Reports in Linguistics 55, 51–82.
Couper-Kuhlen, Elizabeth and Ono, Tsuyoshi. (2007). ‘Incrementing’ in conversation. A com-
parison of practices in English, German, and Japanese. Pragmatics 17(4), 513–552.
Couper-Kuhlen, Elizabeth and Selting, Margaret. (2018). Interactional linguistics: Studying
language in social interaction. Cambridge: Cambridge University Press.
Cristofaro, Sonia. (2016). Routes to insubordination: A cross-linguistic perspective. In Nicholas
Evans & Honoré Watanabe (eds.), Insubordination (pp. 393–422). Amsterdam/Philadel-
phia: John Benjamins. https://doi.org/10.1075/tsl.115.15cri
Deppermann, Arnulf and Günthner, Susanne (Eds.). (2015). Temporality in interaction. Amster-
dam: John Benjamins. https://doi.org/10.1075/slsi.27
Deutscher, Guy. (2009). The Akkadian relative clauses in cross-linguistic perspective. Zeitschrift
für Assyriologie und Vorderasiatische Archäologie 92, 86–105.
https://doi.org/10.1515/zava.2002.92.1.86
Du Bois, John W. (forthcoming). Representing discourse. Linguistics Department, University of
California at Santa Barbara (Fall 2012 version). http://www.linguistics.ucsb.edu/projects/
transcription/representing
Du Bois, John W., Susanna Cumming, Stephan Schuetze-Coburn, and Paolino, Danae. (1992).
Discourse transcription: Santa Barbara papers in linguistics, vol. 4. Santa Barbara: Depart-
ment of Linguistics, University of California, Santa Barbara.
Evans, Nicholas (2007). Insubordination and its uses. In Irina Nicolaeva (Ed.), Finiteness: Theo-
retical and empirical foundations. (pp. 366–431). Oxford: Oxford University Press.
Evans, Nicholas & Watanabe, Honoré. (2016). Insubordination. Amsterdam/Philadelphia: John

Benjamins. https://doi.org/10.1075/tsl.115
Ford, Cecilia E., Fox, Barbara A., & Thompson, Sandra A. (2002). Constituency and the gram-
mar of turn increments. In Cecilia E. Ford, Barbara A. Fox, & Sandra A. Thompson (Eds.),
The language of turn and sequence (pp. 14–38). Oxford University Press, Oxford.
Ford, Cecilia E. & Thompson, Sandra A. (1996). Interactional units in conversation: Syntactic,
intonational, and pragmatic resources for the management of turns. In Elinor Ochs, Eman-
uel A. Schegloff, and Sandra A. Thompson (Eds.), Interaction and grammar (pp. 134–184).
Cambridge, England: Cambridge University Press.
https://doi.org/10.1017/CBO9780511620874.003
Fox, Barbara A. & Thompson, Sandra A. (2007). Relative clauses in English conversation: Rela-
tivizers, frequency, and the notion of construction. Studies in Language 31, 293–326.
https://doi.org/10.1075/sl.31.2.03fox
Glinert, Lewis. (1989). The grammar of Modern Hebrew. Cambridge: Cambridge University
Press.
Günthner, Susanne. (2011). Between emergence and sedimentation: Projecting constructions in
German interactions. In Peter Auer and Stephan Pfänder (Eds.), Constructions: Emerging
and emergent (pp. 156–185). Berlin: Walter de Gruyter.
https://doi.org/10.1515/9783110229080.156
Günthner, Susanne. (2014). The dynamics of dass-constructions in everyday German interac-
tions – a dialogical perspective. In Susanne Günthner, Wolfgang Imo, and Jörg Bücker
(Eds.), Grammar and dialogism (pp. 179–205). Berlin: Walter de Gruyter.
Goodwin, Charles. (1987). Forgetfulness as an interactive resource. Social Psychology Quarterly
50, 115–131. (Special Issue on Language and Society, edited by Douglas Maynard).
https://doi.org/10.2307/2786746
Goodwin, Marjorie Harness & Goodwin, Charles. (1986). Gesture and coparticipation in the
activity of searching for a word. Semiotica 62, 51–75.
Hendery, Rachel. (2012). Relative clauses in time and space: A case study in the methods of dia-
chronic typology. Amsterdam/Philadelphia: John Benjamins. https://doi.org/10.1075/tsl.101
Holmstedt, Robert D. (2007). The etymologies of Hebrew ašer and še. Ancient Near Eastern
Studies 43, 9–28.
Hopper, Paul J. (1987). Emergent grammar. In Jon Aske, Natasha Beery, Laura Michaelis, and
Hana Filip (Eds.), Proceedings of the thirteenth annual meeting of the Berkeley Linguistics
Society 13 (pp. 139–157). Berkeley: Berkeley Linguistics Society.
Hopper, Paul J. (1992). Times of the sign: On temporality in recent linguistics. Time and soci-
ety,1(2), 223–238. https://doi.org/10.1177/0961463X92001002006
Hopper, Paul J. (2001). Grammatical constructions and their discourse origins: Prototype or
family resemblance? In Martin Pütz, Susanne Neimeier, and René Dirven (Eds.), Applied
cognitive linguistics I: Theory and language acquisition (pp. 109–129). Berlin/New York :
Mouton de Gruyter. https://doi.org/10.1515/9783110866247.109
Hopper, Paul J. (2011). Emergent grammar and temporality in interactional linguistics. In
Peter Auer and Stephan Pfänder (Eds.), Constructions: Emerging and emergent (pp. 22–44).
Berlin: Walter de Gruyter. https://doi.org/10.1515/9783110229080.22
Hopper, Paul J. & Thompson, Sandra A. (2008). Projectability and clause combining in interac-
tion. In Ritva Laury (Ed.), Crosslinguistic studies of clause combining: The multifunctionality
of conjunctions (pp. 99–123). Amsterdam: John Benjamins.
https://doi.org/10.1075/tsl.80.06hop
122 Yael Maschler
Hopper, Paul J. and Traugott, Elizabeth Closs. (2003). Grammaticalization. Second edition.
Huehnergard, John. (2006). On the Etymology of the Hebrew Relative. In Steven E. Fassberg
and Avi Hurvitz (Eds.), Biblical Hebrew in its Northwest Semitic setting: Typological and
historical perspectives (pp. 103–125). Jerusalem and Winona Lake: Indiana.
Inbar, Anna. (2016). Is subordination viable? The case of Hebrew ʃɛ ‘that’. Romance Corpora and
Linguistic Studies 3(2), 287–310.
Kärkkäinen, Elise & Sandra A. Thompson. (2018). Language and bodily resources: ‘Response
packages’ in response to polar questions in English. Journal of Pragmatics 123, 220–238.
Keevallik, Leelo. (2008). Conjunction and sequenced action: The Estonian complementizer
and evidential particle et. In Ritva Laury (Ed.), crosslinguistic studies of clause combining:
The multifunctionality of conjunctions (pp. 125–152). Amsterdam/Philadelphia: John Ben-
jamins. https://doi.org/10.1075/tsl.80.07kee
Keevallik, Leelo. (2018). What does embodied interaction tell us about grammar? Research on
Kendon, Adam. (2004). Gesture: Visible action as utterance. Cambridge: Cambridge University
Press. https://doi.org/10.1017/CBO9780511807572
Labov, William. (1972). Language in the Inner City. Philadelphia: University of Pennsylvania
Press.
Laury, Ritva & Seppänen, Eeva-Leena. (2008). Clause combining, interaction, evidentiality,
participation structure, and the conjunction-particle continuum: The Finnish että. In Ritva
Laury (Ed.), Crosslinguistic studies of clause combining. The multifunctionality of conjunc-
tions (pp. 153–178). Amsterdam/Philadelphia: John Benjamins.
https://doi.org/10.1075/tsl.80.08lau
Lehti-Eklund, Hanna. (2002). Om att som diskursmarkör [About that as a discourse marker].
Språk och stil 11, 81–118.
Lerner, Gene. (1991). On the syntax of sentences-in-progress. Language in Society 20, 441–458.
https://doi.org/10.1017/S0047404500016572
Linell, Per. (2013). The dynamics of incrementation in utterance-building: Processes and re-
sources. In Beatrice Szczepek Reed & Geoffrey Raymond (Eds.), Units of talk – units of
action (pp. 57–91). Amsterdam/Philadelphia: John Benjamins.
https://doi.org/10.1075/slsi.25.03lin
Lyons, John. (1968). Introduction to theoretical linguistics. Cambridge: Cambridge University
Press. https://doi.org/10.1017/CBO9781139165570
Maschler, Yael. (2001). veke'ilu haragláyim sh’xa nitka'ot bifním kaze (‘and like your feet get stuck
inside like’): Hebrew kaze (‘like’), ke'ilu (‘like’), and the decline of Israeli dugri (‘direct’)
speech. Discourse Studies 3(3), 295–326. https://doi.org/10.1177/1461445601003003003
Maschler, Yael. (2002). On the grammaticization of ke'ilu (‘like’, lit. ‘as if ’) in Hebrew talk-in-
interaction. Language in Society 31, 243–276. https://doi.org/10.1017/S0047404501020176
Maschler, Yael. (2011). On the emergence of adverbial connectives from Hebrew relative clause
constructions. In Peter Auer and Stephan Pfänder (Eds.), Constructions: Emerging and
emergent (pp. 293–331). Berlin: Walter de Gruyter.
https://doi.org/10.1515/9783110229080.293
Maschler, Yael. (2012). Emergent Projecting constructions: The case of Hebrew yada (‘know’).
Maschler, Yael & Nir, Bracha. (2014). Complementation in linear and dialogic syntax: The case
of Hebrew divergently aligned discourse. Cognitive Linguistics 25(3), 523–557.
https://doi.org/10.1515/cog-2014-0029
Maschler, Yael. (2015). Word order in time: Emergent Hebrew (Ns)V/VNs syntax. In Arnulf
Deppermann and Susanne Günthner (Eds.), Temporality in interaction (pp. 201–236).
Amsterdam: John Benjamins.
Maschler, Yael. (2017). The emergence of Hebrew loydea/loydat (‘I dunno masc/fem’) from
interaction: Blurring the boundaries between discourse marker, pragmatic marker, and
modal particle. In Andrea Sansò and Chiara Fedriani (Eds.), Pragmatic markers, discourse
markers and modal particles: New perspectives (pp. 37–69). Amsterdam/Philadelphia: John
Benjamins. https://doi.org/10.1075/slcs.186.02mas
Maschler, Yael. (2018). The on-line emergence of Hebrew insubordinate she- (‘that/which/who’)
clauses: A usage-based perspective on so-called ‘subordination’. Studies in Language 42(3),
669–707.
Maschler, Yael & Estlein, Roi. (2008). Stance-taking in Hebrew casual conversation via be'emet
(‘really, actually, indeed’, lit. ‘in truth’). Discourse Studies 10(3), 283–316.
https://doi.org/10.1177/1461445608090222
Maschler, Yael & Fishman, Stav. (forthcoming). From multi-clausality to discourse markerhood:
The Hebrew ma she- (‘what that’) construction in so-called ‘pseudo-clefts’.
Maschler, Yael, Polak-Yitzhaki, Hilla, Fishman, Stav, Miller Shapiro, Carmit, Goretsky, Netanel,
Aghion, Gallith, Fofliger, Ophir, Wildner, Nikolaus, & Ben Moshe, Yotam Michael. (2019).
The Haifa Multimodal Corpus of Spoken Hebrew.
Matalon, Nadav. (2016). The Camel Humps prosodic pattern. M.A. thesis, Department of Lin-
guistics, Hebrew University of Jerusalem.
Mertzlufft, Christine & Wide, Camilla. (2013). The on-line emergence of postmodifying att-
and dass-clauses in spoken Swedish and German. In Eva Havu and Irma Hyvärinen (Eds.),
Comparing and contrasting syntactic structures: From dependency to quasi-subordination.
(pp. 199–229). Helsinki: Société Néophilologique.
Mithun, Marianne. (2008). The extension of dependency beyond the sentence. Language 84,
69–119. https://doi.org/10.1353/lan.2008.0054
Mondada, Lorenza. (2016). Challenges of multimodality: Language and the body in social
interaction. Journal of Sociolinguistics 20 (3), 336–366. https://doi.org/10.1111/josl.1_12177
Mondada, Lorenza. (2018). Multiple temporalities of language and body in interaction: Chal-
lenges for transcribing multimodality. Research on Language and Social Interaction 51(1),
85–106. https://doi.org/10.1080/08351813.2018.1413878
Olson, Michael. (1981). Barai clause junctures: Toward a functional theory of interclausal rela-
tions. Ph.D. dissertation, Australian National University.
Pekarek Doehler, Simona. (2011). Clause-combining and the sequencing of actions: Projector
constructions in French talk-in-interaction. In Ritva Laury and Ryoko Suzuki (Eds.), Sub-
ordination in conversation: A cross-linguistic perspective (pp. 103–148). Amsterdam: John
Benjamins. https://doi.org/10.1075/slsi.24.06doe
Pekarek Doehler, Simona, De Stefani, Elwys, & Horlacher, Anne-Sylvie. (2015). Time and
emergence in grammar: Left-dislocation, right-dislocation, topicalization and hanging topic
in French. Amsterdam: John Benjamins. https://doi.org/10.1075/slsi.28
Polak-Yitzhaki, Hilla & Maschler, Yael. (2016). Disclaiming understanding? Hebrew 'ani lo
mevin/a (‘I don’t understand masc/fem’) in everyday conversation. Journal of Pragmatics
124 Yael Maschler
Sacks, Harvey. & Schegloff, Emanuel A. (1979). Two preferences in the organization of refer-
ence to persons and their interaction. In G. Psathas (Ed.), Everyday language: Studies in
ethnomethodology (pp. 15–21). New York: Irvington Publishers.
Sacks, Harvey & Schegloff, Emanuel A. (2002). Home position. Gesture 2(2), 133–146.
https://doi.org/10.1075/gest.2.2.02sac
de Saussure, Ferdinand (1959 [1916]). Course in general linguistics. Edited by Charles Bally and
Albert Sechehaye, in collaboration with Albert Reidlinger. Translated from French by Wade
Baskin. New York: McGraw-Hill.
Schegloff, Emanuel A. (1996). Turn organization: One intersection of grammar and interaction.
In Elinor Ochs, Emanuel A. Schegloff, and Sandra A. Thompson (Eds.), Interaction and
grammar (pp. 52–133). Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511620874.002
Schegloff, Emanuel A. (2001). Conversation Analysis: A project in progress – ‘increments’. Fo-
rum lecture delivered at the LSA Linguistic Institute, University of California, Santa Barbara.
Selting, Margret & Couper-Kuhlen, Elizabeth (Eds.). (2001). Studies in interactional linguistics.
Amsterdam: John Benjamin. https://doi.org/10.1075/sidag.10
Streeck, Jürgen. (2002). Grammars, words, and embodied meanings: On the uses and evolution
of so and like. Journal of Communication 52(3), 581–596.
https://doi.org/10.1111/j.1460-2466.2002.tb02563.x
Streeck, Jürgen. (2009). Gesturecraft: The manu-facture of meaning. Amsterdam: John Benjamins
Publishing Company. https://doi.org/10.1075/gs.2
Streeck, Jürgen. (2013). Praxeology of gesture. In Cornelia Müller, Alan Cienki, Ellen Fricke,
Silva H. Ladewig, David McNeill, and Sedinha Teßendorf (Eds.), Body – language – com-
munication: An international handbook on multimodality in human interaction, Vol. 1.
(pp. 674–688). Berlin: De Gruyter Mouton.
Tao, Hongyin & McCarthy, Michael J. (2001). Understanding non-restrictive which-clauses in
spoken English, which is not an easy thing. Language Sciences 23, 651–677.
https://doi.org/10.1016/S0388-0001(00)00026-7
Tsadka, Yitshak. (1989). taxbir veshixbur [Syntax]. Tel Aviv: Horev.
Van Valin, Robert D. (1984). A typology of syntactic relations in clause linkage. Proceedings of
the tenth annual meeting of the Berkeley Linguistics Society. 542–558.
Verstraete, Jean-Christophe, D’Hertefelt, Sarah, & Van Linden, An (2012). A typology of
complement insubordination in Dutch. Studies in Language 36,123–153.
https://doi.org/10.1075/sl.36.1.04ver
Weinert, Regina. (2012). Complement clauses in spoken German and English: Syntax, deixis
and discourse-pragmatics. Folia Linguistica 46 (1), 233–265.
https://doi.org/10.1515/flin.2012.8
Wide, Camilla. (2014). Constructions as resources in interaction. Syntactically unintegrated att
‘that’-clauses in spoken Swedish. In R. Boogaart, T. Colleman & G. Rutten (Eds.), Expand-
ing the scope of construction grammar (pp. 353–388). Berlin: Mouton de Gruyter.
Wright, Melissa. (2005). Studies of the Phonetics-Interaction Interface: Clicks and Interactional
Structures in English Conversation. PhD dissertation, University of York.
Appendix. Transcription and glossing conventions
Transcription basically follows Chafe (1994), as sometimes adapted by Du Bois et al. (1992) and
Du Bois (forthcoming), with a few additions:
... - pause of 0.1 – 1.0 seconds
. . – perceptible pause of less than 0.1 second
(3.56) – measured pause of 3.56 seconds
, – comma at end of line – continuing intonation (‘more to come’)
. – period at end of line – sentence-final falling intonation
? – question mark at end of line – sentence-final ‘appeal intonation’
?, – question mark followed by comma – ‘continuing appeal’ intonation
! – exclamation mark at end of line – sentence final exclamatory intonation
ø – lack of punctuation at end of line – a fragmentary intonation unit, one which never
reached completion.
two hyphens – elongation of preceding sound
underlined syllable – primary stress of intonation unit
boldfaced syllable – secondary stress of intonation unit
boldfaced words – the construction discussed in the sub-section
@ – a burst of laughter (each additional @ symbol denotes an additional burst)
square bracket to the left of two consecutive lines indicates
beginning of overlapping speech, two speakers talking at once
alignment such that the right of the top line
is placed over the left of the
bottom line indicates latching, no
interturn pause
Musical notation as necessary: f – forte (loud), p – piano (soft), pp – pianissimo (very soft),
dim – diminuendo (progressively softer), acc – accelerando (progressively faster), staccato –
(very short)
[in square brackets] – broad phonetic transcription of preceding utterance.
{in curly brackets} – transcriber’s comments
' – uninverted quotation mark indicates the glottal stop phoneme.
’ – inverted quotation mark in a transliterated word indicates an elided form (e.g., 'an’lo
instead of 'ani lo (lit. ‘I not’, ‘I don’t’ / ‘I’m not’))
- one hyphen – bound-morpheme boundary
/words within slashes/ indicate uncertain transcription
Glossing conventions according to the Leipzig glossing conventions

Transcription of embodied behavior
* * gesture and action descriptions are delimited in between symbols
x x two identical symbols (one symbol per participant and per type of action) are synchro-
nized with corresponding stretches of talk
----> gesture or action described continues across subsequent lines
---->* until the same symbol is reached
Chapter 5
Emergent patterns of predicative clauses

in spoken Hebrew discourse
The ha'emet (hi) she- ‘the truth (is) that’ construction
Hilla Polak-Yitzhaki
This study focuses on the Hebrew construction ha'emet (hi) she- ‘the truth (is)
that’ in a corpus of informal audio and video conversation. Taking an interac-
tional linguistics approach, I argue that the construction serves as a metalingual
fragment constituting a projecting construction. Its employment is fixed and
formulaic, it occurs at a moment of shift in the discourse and is used to project
talk which does three kinds of social work: displaying the speaker’s stance;
setting the record straight regarding the speaker’s personal world; and revealing
delicate information. Each function emerges in a context-sensitive manner,
revealing the relationship between the construction and its sequential position.
Employment of the construction illustrates the ways grammar evolves from the
interactions among conversational participants.
Keywords: projecting construction, Hebrew, predicative clause, routinization
1. Introduction1
In this study I examine Hebrew predicative clauses involving the noun 'emet
‘truth’ – the ha'emet (hi) she- ‘the truth (is) that’ construction – in everyday con-
versation. I outline the kinds of social work this construction may accomplish and
its emergence in interaction.
1. I thank Yael Maschler, Leelo Keevallik, Sandra Thompson, and Elizabeth Couper-Kuhlen for
their constructive and insightful comments and criticism.
https://doi.org/10.1075/slsi.32.05pol
128 Hilla Polak-Yitzhaki
Let us begin with some background about Hebrew predicative clauses. Hebrew
is a language that, aside from verbal clauses, enables also nominal clauses, that is,
clauses which consist of a non-verbal predicate, as in (1)2:
(1) ...ze 'inyan 'axer.
this.m.sg matter different
...this [is] a different matter.
The noun 'inyan ‘matter’ is considered the predicate of this nominal clause.
According to traditional Hebrew grammar (e.g., Nahir, 1987, p. 43; Blau, 1967), a
predicate of a nominal clause may be substituted by another clause, resulting in a
complex syntactic structure, e.g.:
(2) ...ha-maskana she-hu lo dafuk.
def.art-conclusion that-he neg messed_up
...the conclusion [is] that he's not [a] messed up [guy].
The clause shehu lo dafuk ‘that he’s not a messed up guy’ is traditionally analyzed
as a predicative clause, i.e., a clause substituting the predicate of the main clause.
Hebrew predicative clauses are thus considered subordinate clauses of nominal
main clauses. The main clause contains a subject and usually a copula,3 and the
subordinate predicative clause opens with the complementizer she- (‘that’)4 (Blau,
1967). Additional instances of Hebrew predicative clauses are illustrated in (3)–(5):
(3) ..ha-'emetshe-ze nexmad,
the-truth that-this.m.sg nice
..the truth [is] that it’s nice,
(4) ..ve-ha-keta hu--,

and-the-segment he
..and the thing i--s,
..she-hi hayta magniva!
that-she be.pst.3f.sg cool.f.sg
..that she was cool!
2. All examples come from the Haifa Multimodal Corpus of Spoken Hebrew (Maschler et al.,
2019) (see below). Transcription conventions can be found in the Appendix.
3. Since Hebrew has nominal clauses, the copula is optional. This is demonstrated in (2) above,
in which the copula is not employed.
4. Nahir (1987, p. 43) mentions also the complementizer 'asher. In the Haifa Multimodal
Corpus of Spoken Hebrew no tokens of this complementizer were found.
Chapter 5. Emergent patterns of predicative clauses in spoken Hebrew discourse 129
(5) ..ha-be'aya ze she-yesh tsmaxim

the-problem this.m.sg that-there_is plant.pl
'axerim leyado.
other.pl next_to-3m.sg
..the problem is that there are other plants next to it.
All these tokens open with the definite article ha ‘the’ attached to an abstract meta-
phorical noun, such as 'emet ‘truth’, keta ‘(the most important / funny / annoy-
ing) thing’, lit. ‘segment’ (cf. Maschler, 1998), or be'aya ‘problem’, followed by the
complementizer she- ‘that’. Some tokens also contain a copula in past or present
forms before the complementizer.
According to traditional Hebrew grammar, then, the construction ha'emet
(hi) she-[clause] is considered a bi-clausal construction. However, studies of many
constructions in other languages traditionally treated as bi-clausal show that the
part traditionally considered a “main clause” is better described as a formulaic
fragment functioning as a ‘projecting construction’ (Hopper 2004; Auer 2005),5
used to frame the upcoming action.
Schmid (2001), in a study of ‘N-be-that-constructions’, such as the thing is that,
in a corpus of British English classifies the truth is that as an evidential upgrading,
and claims that speakers may use the construction in order to introduce into the
discourse new information with the form of given information.
Edwards and Fasulo (2006) studied English constructions such as to be honest.
They found that these constructions framed the speaker’s stance and were involved
in the speaker’s personal assessments and in dispreferred responses.
Aijmer (2007), who studied the British English construction ‘the N is that’
and its variations in the language, claims that the noun (N) in this construction is
special by being “unclear and abstract” (2007, p. 31). She describes the construc-
tion as formulaic and its function as “rhetorical resources in interaction” (2007,
p. 32), thus, for instance, the fact is that may be used rhetorically to launch a rejec-
tion, and it may also be used to introduce elaborations or to strengthen a position
(2007, p. 44).
Günthner (2008, 2011) studied the German constructions die Sache ist/das
Ding ist ‘the thing is’. She found that the talk following the construction manifests
semantic, syntactic and prosodic properties indicating its independence. There are
thus syntactic reasons for not considering the German phrase a “main clause”.
Hsieh (2018) studied the use of the Mandarin construction wenti-shi ‘(the)
thing is/(the) problem is’ in natural Chinese conversation. The study claims that
5. See also Aijmer 2007; Günthner 2008, 2011; Hopper & Thompson 2008; Maschler 2009,
2012; Pekarek Doehler 2011, 2015; Lindström et al. 2016; Hsieh 2018.
wenti-shi is a projecting construction, which can mark a shift in perspective and be

involved in pursuing agreement in the discourse.
The present study focuses on the ha'emet (hi) she- construction in Hebrew,
a Semitic language, thereby contributing to the comparative study of ‘N-be-that-
constructions’ across languages. The analysis is based entirely on a corpus of
audio- and video-recorded everyday spontaneous interaction (see below) – the
primordial and natural site of language (Schegloff 1993, 1996).
The study reveals that the ha'emet (hi) she- construction is a metalingual frag-
ment whose employment is fixed and formulaic. It accomplishes a shift (Goffman,
1981; Maschler, 1997) in the talk for specific interactional purposes and actions,
and its uses are dependent on its position within the speaker’s turn and within
the ongoing sequence. In the Hebrew corpus the construction was used to project
talk which does three kinds of social work: displaying the speaker’s stance; setting
the record straight regarding the speaker’s personal world; and revealing delicate
information. The sequential position of the construction is correlated with these
different kinds of social work. A functional difference was also found between
tokens manifesting a copula and those with no copula.
In the following section (Section 2) the data are described. Section 3 presents
the structural and functional analysis of the data, and Section 4 discusses these
findings. The study ends with conclusions in Section 5.
2. Data
The study is based on audio and video recordings of spontaneous informal Hebrew
interactions between students, and their friends and relatives. The audio data were
collected during the years 1993–2017, except for one conversation recorded in
1986. This audio corpus consists of 266 conversations, among 770 different in-
terlocutors, comprising over 12 hours of talk (approximately 735 minutes). Each
conversation ranges from 1 to 9 minutes and includes 2–5 participants.
The video data were collected in 2017–2018. They consist of 9 everyday
conversations, among 21 different interlocutors, comprising over 2 hours of talk
(approximately 123 minutes). Each conversation ranges from 5 to 50 minutes and
includes 2–5 participants.
Throughout the corpus 79 cases of predicative clauses were employed, all
represented by the following format:
ha- N (haya/ hayta/ hu/hi/ ze) she- [clause]

the-N (be.pst.m.sg/be.pst.f.sg/he/she/this.m.sg) that-[clause]
the-N (was/is) that-[clause]
Of the 79 predicative clauses found throughout the recordings, 33 tokens

consist of the ha'emet (hi) she- construction. These 33 tokens form the focus of the
present paper.
3. Analysis
3.1 Structural analysis – The ha'emet (hi) she- construction as a discourse

marker
The copula in Hebrew predicative clauses (as in all nominal clauses) is optional.
Even though according to traditional grammar the copula tends to occur in pred-
icative clauses (see introduction), only eleven tokens throughout the corpus (33%)
manifest a copula. Twenty two of the 33 tokens of the ha'emet (hi) she- construction
contain no copula (67%) and are in the form of ha'emet she- ‘the truth (is) that’.
All 33 ha'emet (hi) she- tokens initiate a new TCU and occur at a Transition
Relevance Place or as an interruption of the previous turn. They are all employed in
intonation-unit-initial position. Their position within the turn is shown in Table 1:
Table 1. The position of the ha'emet (hi) she- construction within the turn
Position within the turn Tokens
Turn-initial position 17 52%
Turn-initial position after a DM 9 27%
Turn medially 7 21%
Following a final intonation contour 5 15%
Following a truncated intonation unit 1 3%
Following a DM 1 3%
TOTAL 33 100%
Table 1 reveals that 17 tokens occur in turn-initial position. Another 9 tokens

occur in turn-initial position immediately following a discourse marker. Only
7 tokens occur turn medially, none of them following a continuing intonation
contour (unless they occur after a discourse marker). In this chapter, I will show
what these skewings reveal about the way speakers draw on this construction in
everyday social interactions. Structurally, then, all 33 ha'emet (hi) she- tokens ful-
fill the structural requirement for prototypical discourse markerhood, according
to Maschler (1998, 2009). Maschler (2002) found that 94% of the 613 discourse
markers employed in 16 conversations from the present corpus fulfill the struc-
tural requirement of occurring in intonation-unit initial position. If they occur in
same-speaker talk, they may occur following another discourse marker or follow-
ing any intonation contour other than continuing intonation (see also Maschler
2009, p. 17–20). We can see, then, that none of the ha'emet (hi) she- tokens occur
after a continuing intonation contour unless they occur after another discourse
marker. These prosodic contexts and the position of the construction within the
sequence reveal the strong connection between the construction and Maschler’s
characterization of discourse markers. In fact, other features of discourse markers
also apply to the ha'emet (hi) she- construction: as mentioned at the beginning of
this section, in 67% of the cases the copula is not deployed, and the construction
is shorter: ha'emet she-. Reduction is another feature characterizing discourse
markers (Brinton 1996, p. 33–35; Jucker & Ziv 1998, p. 3–4). Analysis of the
prosodic features of the construction reveals that in 14 cases (42%) there is a
prosodic boundary between the construction and the following clause: in 10 in-
stances speakers employ ha'emet and the optional copula hi in one intonation unit,
whereas the complementizer and the clause are employed in a separate unit; in 4
cases ha'emet and the complementizer she- occur in one intonation unit, and the
projected clause in a separate intonation unit. This prosodic separation from the
clause also contributes to the analysis of the construction as a discourse marker.
A discourse marker, as Maschler (2009, p. 225–229) states, is a project-
ing construction, since it interactionally projects a frame shift. Maschler lists
three differences between projecting constructions (as discussed in Hopper &
Thompson 2008) and discourse markers: a) discourse markers are always meta-
lingual (Maschler 2009, p. 227), b) discourse markers’ projection is more variable
(Maschler 2009, p. 227) and c) discourse markers carry a more specific projec-
tion than projecting constructions that have not (yet) become discourse markers
(Maschler 2009, p. 228). Despite these differences she argues that the phenomenon
discussed in Hopper and Thompson (2008) and the phenomenon discussed in her
study (i.e., discourse markers) share many properties.
Günthner (2011), in her study of the German N-be-that construction, notes the
similarity between the position of the construction – in the pre-front field (Auer,
1996) – and that of discourse markers (among other elements), which also tend to
occupy this position (Günthner 2011, p. 23). In the next sub-section (Section 3.2)
we will see that the Hebrew ha'emet (hi) she- construction is always employed in
the pre-front field, it is always metalingual, and it projects three specific social ac-
tions. We will return to its role as a discourse marker in the discussion (Section 4).
3.2 Functional analysis
A close analysis of all 33 ha'emet (hi) she- tokens reveals that the construction can
be employed to project three main functions:
a. Expressing the speaker’s opinion, stance or evaluation (14 tokens)

b. Setting the record straight regarding the speaker’s personal world (14 tokens)
c. Revealing personal, sensitive or delicate information (5 tokens)
The analysis also reveals that the specific function always emerges in a context-
sensitive manner, that is, there is a motivated relationship between the construc-
tion and its position within both the turn and the sequence.
3.2.1 Expressing the speaker’s opinion, stance or evaluation

In using the terms ‘stance’ and ‘evaluation’ I follow Du Bois (2007), who states
that in taking a stance one can “assign value to objects” (2007, p. 139), position
oneself “with respect to those objects” (ibid.), and express alignment with another
speaker. By ‘object’ Du Bois means the “entity or state of affairs toward which the
speaker expresses their […] stance.” (Du Bois 2007, p. 155). Thus, a stance object
may be a relaxing weekend or a specific place one wishes to visit, and so on. Du
Bois relates the notion of ‘stance’ and ‘evaluation’ by proposing that “[t]he act of
taking a stance necessarily invokes an evaluation at one level or another” (Du Bois
2007, p. 141). He defines evaluation as “the process whereby a stancetaker orients
to an object of stance and characterizes it as having some specific quality or value.”
(Du Bois 2007, p. 143).
In the Hebrew corpus speakers expressed, via the ha'emet (hi) she-construction,
their evaluation regarding some stance object. For example, in the following audio
excerpt two female friends, Dana and Lilach, discuss the forthcoming wedding of
Ayelet, another female friend of theirs:
(6) ‘Engagement and betrayal’
145 DAN: ..nifgasht 'im 'ayelet kvar?
meet_with.pst.2f.sg with Ayelet already
..have you already met with Ayelet?
146 LIL: ..ke--n!

..ye--s!
147 ..yashavnu le-sushi--,

sit.pst.1pl to-sushi
..we sat down for sushi--,
148 ...shatinu yayi--n,

drink.pst.1pl wine
...we drank some wi--ne,
149 DAN: ..lo 'amart li.

not say.pst.2f.sg to-1sg
..you didn’t tell me.
150 LIL: ...'asinu mamash mesibat 'irusin.

do.pst.1pl really party.of engagement
...we had a real engagement party.
151 ....ken,
....yes,
152 ..ra'iti ta’- taba'at!
see.pst.1sg acc-the-ring
..I saw the ring!
153 DAN: ...'oy va'avoy.
...oh no.
154 LIL: ...lama?
...why?
➔ 155 ....ha-'emet she-ze lo kaze--,
the-truth that-this.m.sg not such
....the truth [is] that it’s not so--,
156 ..ke'ilu--,
..li--ke,
157 ..faka--tsi,
..tra--shy,
158 ..ha-tixnun xatuna shela,
the-planning.of wedding of-her
..her wedding plan,
159 ..kmo she-xasha
like that-think.pst.
..as thou[ght?]
160 kmo she-'anaxnu betax xoshvot.
like that-we probably think.prs.f.pl
as we probably think.
161 ..nir'a li.
seem.prs.m.sg to-1sg
..it seems to me.
Lilach shares with Dana her excitement about celebrating Ayelet’s engagement:
..yashavnu lesushi--, …shatinu yayi--n, ‘..we sat down for sushi--, …we drank some
wi--ne,’ (6. 147–148) …'asinu mamash mesibat 'irusin. ‘…we had a real engagement
party.’ (6. 150). When she mentions that she saw the ring (6. 152), Dana responds
with …'oy va'avoy. ‘…oh no.’ (6. 153). This negative stance is employed with no
account (see Ford 2001a, 2001b), not even after Lilach’s explicit request for one:
…lama? ‘…why?’ (6. 154). Lilach, then, following a final intonation contour and
a rather long pause of 1.1 seconds, self-selects to be the next speaker and presents
her own stance regarding Ayelet’s upcoming wedding: ….ha'emet sheze lo kaze--,
..ke'ilu--, ..faka--tsi, ..hatixnun xatuna shela, ‘….the truth [is] that it’s not so--, ..li-
-ke, ..tra--shy, ..her wedding plan,’ (6. 155–158). Lilach introduces this stance with
the ha'emet she- construction with no copula, employed at intonation-unit-initial
position at a TRP within her own turn. It serves as a response to the interlocutor’s
expressed stance (6. 153) and disaffiliates (Stivers, 2008) with it: the anticipated
wedding will not be as trashy as they probably think. The construction signals the
shift from Lilach’s relating to Dana’s evaluation of Lilach’s having met Ayelet to
presenting Lilach’s own stance regarding both the wedding and Dana’s negative
stance towards it. The construction, then, projects here a reaction to Dana’s stance
by explaining Lilach’s own stance.
The following audio excerpt illustrates an instance of the ha'emet hi she- con-
struction in which the speaker agrees with the recipient. In this conversation two
female friends are discussing the nature of the Hebrew language:
(7) ‘Reduplication Stress in Hebrew and Russian’
362 LIA: ..ze ha-koshi,
this.m.sg the-difficulty
..this is the difficulty,
(three intervening intonation units)
366 ...she-yesh hevdel me'od gado--l,

that-there_is difference very big
...that there is a big differe--nce,
367 ...ben ha--,

between the
...between,
(three intervening intonation units)
371 dibu--r?,
speaki--ng?,
372 ...le-ven ha-kti--v.

to-between the-spelling
...and spelli--ng.
➔ 373 KAR: ...ha-'emet hi she-ken.

the-truth she that-yes
...the truth is that there is.
374 yesh hevdel,

there_is difference
there is a difference,
375 ..naxon.
right
..that’s right.
Liat claims that there is a very big difference in Hebrew between the spoken lan-
guage and its spelling (7. 366–367, 371–372), and Karina agrees with her. Karina’s
evaluation, …ha'emet hi sheken. yesh hevdel, ..naxon. ‘…the truth is that there is.
there is a difference, that’s right.’ (7. 373–375), opens with the ha'emet hi she- con-
struction, aligning with Liat’s stance.
The ha'emet hi she- construction is produced here immediately following Liat’s
turn, after a short pause of 0.3 seconds, as opposed to the longer pause of 1.1 sec-
onds in the previous excerpt (6. 155) before the ha'emet (hi) she- construction.
We can see, then, that the ha'emet (hi) she- construction framing a disaffilative
response is produced after a longer pause than the ha'emet hi she- construction
aligning with the interlocutor’s stance.
The data reveal a correlation between the work of the ha'emet (hi) she- con-
struction and its position within its turn and its sequence here: tokens agreeing
with a previous speaker are employed either immediately following the addressee’s
turn (see Example 7 above) or overlapping with it. Tokens presenting disagree-
ment with no explanation always occur immediately following the addressee’s
turn after a discourse marker, and tokens expressing disagreement followed by an
explanation occur turn medially, as illustrated by excerpt (6) above. We will return
to this point later.
The following excerpt is taken from a video interaction between three
friends: Maya, the host, and Hadas and Galia, who are visiting her. Maya tells
Hadas and Galia about her upcoming trip to Africa with three friends of hers
who are all interested in Uganda. She specifies the different interests of each of
them, saying that one is focusing on literature, the other on religion, and the third
on waste management:
(8) ‘Uganda’
130 MAY: weyst menedʒment,
waste management,
131 ..kaze?,
..like?,
132 GAL: hm!
133 MAY: ...ve-'ani be-xinux,

and-I in-education
...and I [focus] on education,
134 ..ve-kaze 'arba'atenu,

and-like four.of-us
..and like the four of us,
135 kaze xoshvim,

like think.PRS.M.PL
like are thinking,
136 la'asok be-'uga--nda,

deal.INF in-Uganda
of focusing on Uga--nda,
137 'o she-kvar

or that-already
or are already
138 ..(inhale)
139 ..mit'askim,
deal.PRS.M.PL
..focusing [on it],
140 'az kaze--,

so li--ke,
141 GAL: ...ma'ayan magi'a la-'arets.

Maayan arrive.PRS.M.SG to-the-country
...Maayan is coming to the country.
142 MAY: ..'az ze nexmad.

so this.M.SG nice
..so it's nice.
143 HAD: 'ah,

oh,
144 GAL: weyst menedʒment.

waste management.
145 HAD: ken?

yes?
146 GAL: ..'im medabrim 'al weyst menedʒment.
if speak.PRS.M.PL on waste management
..speaking of waste management.
147 HAD: @@@@
148 MAY: ha-'emet she-

the-truth that-
'ulay keday la-hem lehakir.

maybe worthwhile to-them know.INF
the truth [is] that maybe they should meet.
149 GAL: ...mi?

...who?
150 MAY: ...ke'ilu le-ma'ayan ve--,

like to-Maayan and,
...like Maayan a--nd,
'ulay keday la-hem lehakir.
maybe worthwhile to-them know.INF
the truth [is] that maybe they should meet.
138 Hilla
149 Polak-Yitzhaki
GAL: ...mi?
...who?
150 MAY: ...ke'ilu le-ma'ayan ve--,

like to-Maayan and,
...like Maayan a--nd,
151 miryam.
Miriam.
152 ...mi she-nosa'at 'itanu.

who that-ride.PRS.F.SG with-us
the one who’s going with us.
153 ....hi mamash xamuda.

she really cute.F.SG
....she's really cute.
154 GAL: ...nakir beynehem.

introduce.FUT.1PL between-them
...we'll introduce them.
155 MAY: ...hi 'ovedet ba-moshav,

she work.PRS.F.SG in-the-Moshav
...she works in the Moshav,
Maya mentions that one of her friends focuses on waste management: weyst
menedʒment, ..kaze? ‘waste management, ..like?,’ (8. 130–131) and this detail
reminds Galia of Maayan, a mutual acquaintance of the three participants, who
is apparently also interested in waste management (8. 146). When Galia mentions
that Maayan is about to visit the country, she addresses Hadas, who displays her
interest in that detail ('ah, ‘oh,’ (8. 143) ken? ‘yes?’ (8. 145)). At that point Maya
takes the turn and with minimal overlap with Hadas’s laughter (8. 147) she initiates
a turn with the ha'emet she- construction: ha'emet she'ulay keday lahem lehakir.
‘the truth [is] that maybe they should meet’, (8. 148). This construction introduces
Maya’s personal stance and shifts the talk away from Maayan’s upcoming arrival
to an assertion that Miriam and Maayan should meet.
Figure 1. Maya gazes at Galia and Hadas

And indeed, it can be seen that Hadas and Galia, who are gazing at each other in
lines 143–147 (Figure no. 1), immediately turn their gazes towards Maya in line
148 (Figure no. 2).
Figure 2. Galia and Hadas gaze at Maya
3.2.2 Setting the record straight regarding the speaker’s personal world
Another 14 tokens introduced by the ha'emet (hi) she construction can be charac-
terized as functioning to set the record straight regarding the speaker’s personal
world. Examine, for instance, excerpt (9), in which Yifat reminds her friend Topaz
of the television series ‘The Moomins’:
(9) ‘The bride and the Groke’
71 YIF: (2.6) 'ata zo
you.M.SG reme??.PRS
(2.6) do you reme[mber?]
72 ../'a/
../oh/
73 ..zoxer 'et ha-muminim?

remember.PRS.M.SG ACC the-Moomin.PL
..do you remember the Moomins?
74 ..ta’- toxnit ha-muminim?

ACC-the-program the-Moomin.PL
..the television series the Moomins?
75 TOP: be-vaday.
in-certain
of course.
(21 intervening intonation units)

(21 intervening intonation units)
97 TOP: ...be-kitsur 'az ma,

in-short so what
...in short so what,
98 ..ma legabey ha-muminim.

what about the-Moomin.PL
..what about the Moomins.
99 YIF: ...haya 'et ha-grok.

be.PST.3M.SG ACC the-Groke
...there was the Groke.
100 ...'ata zoxer?

you.M.SG remember.PRS.M.SG
...do you remember?
101 TOP: ...ha-grok hu me'od mafxid.

the-Groke he very scary
...the Groke
...the Groke is very
is very scary.
scary.
102 ...'ani ve-'eyal,

I and-Eyal
...I and Eyal,
103 YIF: 'az yesh lanu pesel,

so there_is to-1PL statue
so we have a statue,
104 ..she-nir'a mamash kmo @ha-grok,

that-look.3M.SG really like the-Groke
..that looks exactly like @the @Groke,
105 ..leyad ha-me'onot.

near the-dorm.PL
..near the dorms.
106 ..'al ha-deshe.

on the-grass
..on the grass.
107 TOP: ha-'emet she-'an-lo zoxer.. bixlal 'ex hu nir'a,

the-truth that-I-not remember. at_all how he look.
PRS.M.SG PRS.M.SG.
the truth [is] that I don’t remember.. at all what
it looked like,
108 'ani rak zoxer she-hu mamash mamash mafxid.

I only remember.PRS.M.SG that-he really really scary
I only remember it was really really scary.
Topaz confirms that he remembers the Moomins (9. 75) and asks Yifat to continue
her story about them: ..ma legabey hamuminim. ‘..what about the Moomins.’ (9.
98). Yifat mentions the Groke, and asks whether Topaz remembers it as well (9.
99–100). It seems that Yifat interprets Topaz’s answer …hagrok hu me'od mafxid.
‘…the Groke is very scary’ (9. 101) as confirming this memory, because she contin-
ues her story about the Groke and tells about the statue near the dorms (9. 103–105).
At this point Topaz initiates a turn with ha'emet she- to inform Yifat that
he doesn’t remember what the Groke6 looked like (9. 107). Not only does this
construction index a shift from Yifat’s story to Topaz’s current lack of access to
the referent, it also sets the record straight on his previous comment from line 101
‘…the Groke is very scary.’, clarifying that except for the Groke being very scary, he
remembers nothing about it.
As we saw with the previous category of expressing stance and evaluation,
in these cases of setting the record straight the construction also accomplishes a
“break” from the prior talk.
In (9) we saw a speaker setting the record straight with respect to his own
prior assertion. But speakers may also set the record straight regarding an inter-
locutor’s utterances. In the following excerpt, taken from an audio conversation
between two friends, Gad and Yaron, Gad suggests in a jocular way how Yaron
could secretly sleep at his workplace:
(10) ‘Slicer’
46 GAD: ...shulxan kaze gadol ba-misrad?
desk like-this big in-the-office
...such a big desk at the office?
47 YAR: ...'e--,
...uh--,
48 ..dey gadol.
..pretty big.
49 ..ken.
..yes.
50 GAD: .../ve-mitaxat/ la--,

/and-underneath/ to-the
.../and underneath/ the--,
51 mitaxat la
underneath to-the
underneath the
6. Topaz employs the 3rd person singular pronoun hu ‘it’ (lit. ‘he’) in line 107 and in line 108
to refer to the Groke. It is clear that these pronons do not refer to the statue Yifat mentions in
line 103, since the exsistence of the statue is new to Topaz, and since lines 107–108 tie back to
Topaz’s utterance from line 101 …hagrok hu me'od mafxid. ‘…the Groke is very scary’, which
explicitly refer to the Groke.
52 ba- 'emtsa kaze?,

in-the-middle like
like in the middle?,
53 ...lasim kaze,
put.inf like
...to put like,
54 mizro--n katan,
mattress small
a small mattre--ss,
55 ve-kari--t?,
and-pillow
and a pillo--w?,
56 YAR: @@@@@.. @@
57 GAD: ve-'az ta-holex kaze,

and-then you-go.prs.m.sg like
and then you like go,
58 lisho--n kaze,
sleep.inf like
like to slee--p,
59 ..mitaxat la--,
underneath to-the
..underneath the--,
➔ 60 YAR: ....ha-'emet,
the-truth
....truth [is],
61 she- ba- mishmarot 'erev,

that-in-the-shifts.of evening
that in the evening shifts,
62 'en pashut 'anashim.

there_isn’t simple people
there are simply no people.
63 ..'az 'efshar lishon 'al 'eyze--,
so possible sleep.inf on some
..so it’s possible to sleep on so--me,
64 ...sapa sham.
...sofa there.
In the beginning of this excerpt Gad asks whether Yaron has ‘…such a big desk at
the office?’ …shulxan kaze gadol bamisrad? (10. 46), He suggests that Yaron could
sleep under his desk so that nobody would see him (10. 50–55, 57–59), but Yaron
initiates his turn with ha'emet she- to set straight Gad’s misconception: since no
one would be at his office, he could use a sofa and needn’t hide in order to sleep
at his workplace.
The projective capacity of the ha'emet she- construction makes it possible to
use it for extended turns, where turn units can continue building on the already
emerging structure. Yaron continues to rely on the projective capacity of the
construction, and following completion a clause at line 62 with a final intonation
contour 'en pashut 'anashim. ‘there are simply no people.’, he self-selects to be the
next speaker and continues with another clause, 'az 'efshar lishon 'al 'eyze--, …sapa
sham. ‘so it’s possible to sleep on so--me, sofa there,’ (10. 63–64). There are, then,
two clauses here working to set the record straight regarding the interlocutor’s
utterances, both of which are introduced with the ha'emet she- construction.
In these five cases of setting the record straight by relating to the addressee’s
previous utterances, the speaker takes a turn to counter a co-participant’s miscon-
ception.
3.2.3 Revealing personal, sensitive or delicate information

Five tokens of the ha'emet (hi) she- construction in the data serve to reveal personal
or delicate information. Consider the following excerpt, in which Shirley tells her
friend Noa that she and her boyfriend are planning to move in together. Here she
reminds Noa that in the past Noa had offered to arrange for her to consult about a
mortgage with her father, who is a banker:
(11) ‘Hot Seat’
4 NOA: ..ve-lagur levad?
and-live.inf alone
..and to live alone?
5 SHI: ...'e--,
...u--h,
➔ 6 ....ha-'emet hi she-'anaxnu xoshvim,

the-truth she that-we think.prs.m.pl
....the truth is that we’re thinking,
7 lavo ledaber 'im 'aba shelax.

come.inf speak.inf with father yours
of coming to speak to your father.
8 ...lir'ot ma--,
see.inf what
...to see wha--t,
9 'az 'im ha--,

then with the--,
10 ..she-'amart lanu 'im ha-mashkanta,

that-say.pst.2f.sg to-1pl with the-mortgage
..that you told us regarding the mortgage,
11 ve-ze,
and-this.m.sg
and stuff,
12 she-ke'ilu,
that-like
that like,
13 hu yaxol la'azor lanu?

he can.prs.m.sg help.inf to-1pl
he could help us?
14 NOA: 'atem muzmanim.

you.m.pl invited.prs.m.pl
you’re invited.
Although Noa’s question occurs at the beginning of the recorded conversation, it

is still clear that she is asking Shirley about her intention to move away from home:
..velagur levad? ‘..and to live alone?’ (11. 4). Shirley’s response about speaking with
Noa’s father (11. 6–13) does not constitute a direct answer to Noa’s question, but
indirectly suggests that Noa might arrange for her and for her boyfriend to meet
with Noa’s father, to advise them about a mortgage. This is a sensitive moment in
the discourse: Shirley asks here for Noa’s help concerning financial matters. The
ha'emet hi she- construction signals a shift in the talk from supplying an answer to
Noa’s question to a request that introduces delicate information.
Here the talk following the ha'emet hi she- construction is spread across eight
intonation units, displaying the strong projective capacity of this construction (cf.
Günthner 2011, p. 24–27), and it is full of repetitions, hesitations, elongations, and
increments revealing Shirley’s discomfort at her request.
All five tokens revealing personal or delicate information, then, mark a shift
in the talk. This is reminiscent of the ways English well in turn-initial position
functions when it serves as a response to wh- questions (Schegloff & Lerner 2009,
particularly pp. 94–95): it alerts the participants “that the response will be in some
respect not straightforward” (Schegloff & Lerner 2009, p. 101). Here, Shirley’s
response is certainly not straightforward, and the ha'emet hi she- construction at
the beginning of her disaligning turn serves as an alert to Noa that her response is
not going to be straightforward. All tokens show a correlation between the work
they are doing and their sequential position: they are all employed as responsive
to a just-prior turn.
4. Discussion
All 33 ha'emet (hi) she- tokens in the data occur at a TRP or overlapping the previ-
ous turn. Even though the construction consists of a complementizer, suggesting a
relation of subordination, the ha'emet (hi) she- tokens do not display a normatively
bi-clausal pattern: the talk following the ha'emet (hi) she- construction tends to
be longer than a clause, and the ha'emet (hi) she- part does not hold “the relevant
information for the following discourse” (Günthner 2011, p. 14); it does not
convey the “main event” (ibid), as Günthner reports for the N be-part in spoken
German. It therefore is not best seen as a “main clause”, but rather as a projecting
construction foreshadowing the upcoming action.
Since none of the ha'emet (hi) she- tokens occur following a continuing into-
nation contour (unless they occur as part of a cluster of discourse markers), they
all fulfill the structural requirement for being considered a prototypical discourse
marker (Maschler 1998, p. 31, 2009, p. 17). All ha'emet (hi) she- tokens also fulfill
the semantic requirement for being considered a discourse marker (ibid.): they are
all metalingual fragments, since, with the semantically non-objective word 'emet
‘truth’, they refer to the world of the interaction rather than to the extra-lingual
world. Furthermore, most of them display morphosyntactic reduction, as 67% of
the cases do not manifest a copula. There is also a tendency for the ha'emet (hi)
she- part to occur in a separate intonation unit, and even when it does not con-
stitute an intonation unit of its own, it always occupies the intonation unit initial
position, a position often occupied by discourse markers. However, throughout
this paper we have seen evidence of the projective capacity of the construction.
This raises the question of whether the ha'emet (hi) she- construction should be
considered a discourse marker or a projecting construction, and indeed, is there a
difference between projecting constructions and discourse markers?
Maschler draws the connection between projecting constructions and dis-
course markers stating that “[a] discourse marker is a projecting construction”
(2009, p. 229), but one whose projecting possibilities are narrower. The findings
of the present study support this claim. The projecting possibilities of the ha'emet
(hi) she- construction are rather narrow; throughout the corpus we find it project-
ing one of only three main actions: expressing the speaker’s opinion, stance or
evaluation, setting the record straight regarding the speaker’s personal world, or
revealing personal, sensitive, or delicate information.
All ha'emet (hi) she- tokens shift the discussion to the speaker’s personal
experiences, suggestions, ideas, or stances or to revealing the speaker’s personal
information. We have seen how the three functions of ha'emet (hi) she- described
in Section 3.2 are managed by speakers in everyday Hebrew conversation: The
construction may project objection or agreement, evaluation or conviction (which
are all related to expressing the speaker’s stance), a clarification or an account

(related to setting the record straight), a request or a proposal (related to posing
delicate information). Thus, it always signals a break in the ongoing line of talk,
and in this way it is disaligning with respect to the immediately prior action.
In using the ha'emet (hi) she- construction, which contains the word 'emet
‘truth’, speakers claim to possess “the truth” about a certain matter, and establish
themselves as an epistemic authority on what they are about to say. Speakers
obviously have epistemic authority when it comes to their own private domain. It
seems, then, that there is a semantic motivation for the fact that the ha'emet (hi)
she- construction projects talk concerning the speakers’ personal world.7
The analysis reveals a correlation between these three functions and the posi-
tion of the construction within the turn, as illustrated in Table 2:
Table 2. The correlation between the function of the ha'emet (hi) she- construction and
its sequential position
Position of construction within the turn Function implemented
Turn initial position – Introducing delicate information
– Evaluation agreeing with the interlocutor
Turn initial position following a DM – Evaluation disagreeing with the interlocutor
– Setting the record straight regarding the
interlocutor’s utterances
Turn medially – Evaluation and elaboration
– Setting the record straight regarding the
speaker’s utterance
When the speaker threatens the addressee’s face (Brown & Levinson 1987) – by
expressing evaluations disaligning with the interlocutor or by setting the record
straight regarding the interlocutor’s utterances – the construction tends to be
preceded by a discourse marker, thus delaying the response.
When the construction is employed turn medially, the speaker tends to elabo-
rate on his or her own evaluations or to set the record straight regarding his or her
previous utterances.
5. Conclusion
Thirty-three tokens of the ha'emet (hi) she- construction occur in my Hebrew

conversation corpus, none of them in past tense, indicating that this is a fixed and
7. I thank Leelo Keevallik for suggesting this idea to me.

formulaic construction. The ha'emet (hi) she- tokens are metalingual fragments,
and work to shift the talk to the speaker’s subjective personal world: the speaker’s
experiences, ideas, or stances.
We have seen that the ha'emet (hi) she- fragment does not function as a “main
clause”, and that the talk following the ha'emet (hi) she- part tends to be longer
than a clause, so the ha'emet (hi) she- construction cannot be viewed as a main
clause with a subordinate clause, as suggested by traditional grammar, but rather
should be treated as a projecting construction. In this way, employment of the
construction illustrates the ways grammar is shaped by the social interactions
among interlocutors.
References
Aijmer, K. (2007). The interface between discourse and grammar: the fact is that. In A. Celle,
& R. Huart (Eds.), Connectives as Discourse Markers (pp. 31–46). Amsterdam: John Ben-
jamins. https://doi.org/10.1075/pbns.161.05aij
Auer, P. (1996). The pre-front field position in spoken German and its relevance as a gram-
maticalization position. Pragmatics 6(3), 295–322. https://doi.org/10.1075/prag.6.3.03aue
Auer, P. (2005). Projection in interaction and projection in grammar. Text 25(1), 7–36.
Blau, Y. (1967). yesodot hataxbir [The Foundations of Syntax] (in Hebrew). Jerusalem: The
Hebrew Institution for literal Education in Israel.
Brinton, L. J. (1996). Pragmatic Markers in English: Grammaticalization and Discourse Func-
tions. Berlin: Mouton de Gruyter. https://doi.org/10.1515/9783110907582
Brown, P. & Levinson, S. (1987 [1978]). Politeness: Some Universals in Language Usage. Cam-
Chafe, W. (1994). Discourse, Consciousness, and Time: The Flow and Displacement of Conscious
Experience in Speaking and Writing. Chicago: University of Chicago.
Du Bois, J. W. (2007). The stance triangle. In R. Englebretson (Ed.), Stancetaking in Discourse:
Subjectivity, Evaluation, Interaction (pp. 139–182). Amsterdam: John Benjamins.
https://doi.org/10.1075/pbns.164.07du
Du Bois, J. W. (forthcoming). Representing discourse. Linguistics Department, University of
California at Santa Barbara (Fall 2012 version). http://www.linguistics.ucsb.edu/projects/
transcription/representing
Du Bois, J. W., Cumming, S., Schuetze-Coburn S., and Danae P. (1992). Discourse Transcrip-
tion: Santa Barbara Papers in Linguistics, vol. 4. Santa Barbara: Department of Linguistics,
University of California, Santa Barbara.
Edwards, D. & Fasulo, A. (2006). “To Be Honest”: Sequential Uses of Honesty Phrases in Talk-
in-Interaction. Research on Language and Social Interaction 39(4), 343–376.
Ford, C. E. (2001a). Denial and the construction of conversational turns. In J. Bybee, & M.
Noonan (Eds.), Complex Sentences in Grammar and Discourse (pp. 61–78). Amsterdam:
John Benjamins.
Ford, C. E. (2001b). At the intersection of turn and sequence: negation and what comes next.
In M. Selting, & E. Couper-Kuhlen (Eds.), Studies in interactional linguistics (pp. 51–79).
Amsterdam: John Benjamins. https://doi.org/10.1075/sidag.10.05for
Goffman, E. (1981). Forms of Talk. Philadelphia: University of Pennsylvania Press.
Günthner, S. (2008). ‘Die Sache ist …’: eine Projektorkonstruktion im gesprochenen Deutsch.
Zeitschrift für Sprachwissenschaft 27(1), 39–72. https://doi.org/10.1515/ZFSW.2008.003
Günthner, S. (2011). N be that-constructions in everyday German conversation. In R. Laury, &
R. Suzuki (Eds.), Subordination in Conversation: A Crosslinguistic Perspective (pp. 11–36).
Amsterdam: John Benjamins. https://doi.org/10.1075/slsi.24.03gun
Hopper, P. J. (2004). The openness of grammatical constructions. Papers from the 40th Regional
Meeting of the Chicago Linguistic Society 40, 153–175.
Hopper, P. J. & Thompson, S. A. (2008). Projectability and clause combining in interaction. In R.
Laury (Ed.), Crosslinguistic Studies of Clause Combining: The Multifunctionality of Conjunc-
Hsieh, C. Y. C. (2018). From turn-taking to stance-taking: Wenti-shi ‘(the) thing is’ as a projector
construction and an epistemic marker in Mandarin conversation. Journal of Pragmatics
Jucker, A. H. & Ziv Y. (1998). Discourse markers: Introduction. In A. H. Jucker, & Y. Ziv (Eds.),
Discourse Markers: Descriptions and Theory (pp. 1–12). Amsterdam: John Benjamins.
https://doi.org/10.1075/pbns.57.03juc
Lindström, J., Maschler, Y. and Pekarek Doehler, S. (2016). A cross-linguistic perspective on
grammar and negative epistemics in talk-in-interaction. Journal of Pragmatics 106, 72–79.
Maschler, Y. (1997). Discourse markers at frame shifts in Israeli Hebrew talk-in-interaction.
Pragmatics 72, 183–211. https://doi.org/10.1075/prag.7.2.04mas
Maschler, Y. (1998). rotse lishmoa keta? ‘wanna hear something weird/funny [lit. ‘a segment’]?’:
The discourse markers segmenting Israeli Hebrew talk-in-interaction. In A. H. Jucker, &
Y. Ziv (Eds.), Discourse Markers: Descriptions and Theory (pp. 13–59). Amsterdam: John
Benjamins. https://doi.org/10.1075/pbns.57.04mas
Maschler, Y. (2002). The role of discourse markers in the construction of multivocality in Israeli
Hebrew talk in interaction. Research on Language and Social Interaction 35(1), 1–38.
https://doi.org/10.1207/S15327973RLSI35-1_1
Maschler, Y. (2009). Metalanguage in Interaction: Hebrew Discourse Markers. Amsterdam, Phila-
delphia: John Benjamins. https://doi.org/10.1075/pbns.181
Maschler, Yael, Polak-Yitzhaki, Hilla, Fishman, Stav, Miller Shapiro, Carmit, Goretsky, Netanel,
Aghion, Gallith, Fofliger, Ophir, Wildner, Nikolaus, & Ben Moshe, Yotam Michael. (2019).
The Haifa Multimodal Corpus of Spoken Hebrew.
Nahir, S. (1987). 'ikarey torat hamishpat [The Essence of the Theory of the Sentence] (in Hebrew).
Haifa: The Re'ali school publication.
structions in French talk-in-interaction. In R. Laury, & R. Suzuki (Eds.), Subordination in
Conversation: A Cross-Linguistic Perspective (pp. 103–148). Amsterdam: John Benjamins.
Pekarek Doehler, S. (2015). Grammar, projection and turn-organization il y a NP ‘there is NP’

as project construction in French talk-in-interaction. In A. Deppermann, & S. Günthner
(Eds.), Temporality in Interaction (pp. 173–200). Amsterdam/Philadelphia: John Benjamins
Publishing Company.
Schegloff, E. A. (1993). Reflections on quantification in the study of conversation. Research on
Language and Social Interaction 26, 99–128. https://doi.org/10.1207/s15327973rlsi2601_5
Ochs, E. A. Schegloff, and S. A. Thompson (Eds.), Interaction and Grammar (pp. 52–133).
Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511620874.002
Schegloff, E. A. & Lerner, G. H. (2009). Beginning to Respond: Well-Prefaced Responses to Wh-
Questions. Research on Language and Social Interaction 42(2), 91–115.
https://doi.org/10.1080/08351810902864511
Schmid, H. J. (2001). ‘Presupposition can be a bluff ’: How abstract nouns can be used as presup-
position triggers. Journal of Pragmatics 33(10), 1529–1552.
https://doi.org/10.1016/S0378-2166(01)00027-3
Stivers, T. (2008). Stance, alignment and affiliation during story telling: When nodding is a
token of preliminary affiliation. Research on Language in Social Interaction 41, 29–55.
https://doi.org/10.1080/08351810701691123
Appendix. Transcription conventions
Each line denotes an intonation unit (Chafe, 1994) and is followed by an English gloss. Where
this gloss is not close enough to an English utterance, it is followed by a third line supplying a
usually literal translation. Transcription basically follows Chafe 1994, as sometimes adapted by
Du Bois et al. 1992 and Du Bois (forthcoming), with a few additions:
... - pause of 0.1 – 1.0 seconds
. . – perceptible pause of less than 0.1 seconds
(3.56) – measured pause of 3.56 seconds
, − comma at end of line – continuing intonation (‘more to come’)
. – period at end of line – sentence-final falling intonation
? – question mark at end of line – sentence-final ‘appeal intonation’
(Du Bois et al., 1992).
?, − question mark followed by comma – ‘continuing appeal’ intonation (Du Bois, forthcom-
ing)
! – exclamation mark at end of line – sentence final exclamatory intonation
– lack of punctuation at end of line – a fragmentary intonation unit, one which never
reached completion.
-- two hyphens – elongation of preceding vowel sound or consonant
underlined syllable – primary stress of intonation unit
@ – a burst of laughter (each additional @ symbol denotes an additional burst)
square bracket to the left of two consecutive lines indicates

beginning of overlapping speech, two speakers talking at once
alignment such that the right of the top line
is placed over the left of the

bottom line indicates latching, no
interturn pause
pp – pianissimo (spoken very softly)
ff–- fortissimo (spoken very loudly)
/??????/ – transcription impossible
/words within slashes/ indicate uncertain transcription
{in curly brackets} – transcriber’s comments
[xxxxx] – material within square brackets in the gloss indicates exuberances of translation
(what is not there in the original).
'– uninverted quotation mark in the middle of a transliterated word indicates the glottal stop
phoneme.
’ – inverted quotation mark in the middle of a transliterated word indicates an elided form
(e.g., ‘an’lo instead of 'ani lo (lit. ‘I not’, ‘I don’t’ / ‘I’m not’)).
- one hyphen – bound-morpheme boundary
m – masculine
f – feminine
sg – singular
pl – plural
acc – definite accusative marker
cop – copula
Chapter 6
From matrix clause to turn expansion

The emergence of wo juede ‘I feel/think’ in
Mandarin conversational interaction
Wei Wang and Hongyin Tao
One way complex clauses manifest themselves is through a combination of a ma-

trix clause and a complement clause. However, matrix clauses as represented by
constructions such as I think have been widely reported to undergo grammatici-
zation, whereby they become a marker indicating the speaker’s epistemic stance.
This has also been identified for Mandarin Chinese. In this chapter, however,
we report that wo juede ‘I feel/think’ in Mandarin has developed a conversation
interactional function that extends turns-at-talk – a phenomenon that has not
yet received much attention in the Chinese literature on talk in interaction. By
providing an account for the link between the epistemic/evaluative uses and the
turn-expansion function of the erstwhile matrix clause, this paper brings in an
extended, interactional dimension to the study of clause-combining.
Keywords: matrix clause, Mandarin Chinese, turn-expansion, complement

clause
1. Introduction1
One way complex clauses manifest themselves is through a combination of a ma-

trix clause and a complement clause. Matrix clauses in conversational discourse
often appear as a first person singular pronoun plus a cognitive verb exemplified
by I think in English (Thompson & Mulac 1991a, 1991b; Scheibman 2001). The
Mandarin Chinese equivalent is wo juede. Traditionally, the Mandarin verb juede
1. We thank Sandy Thompson and the editors of this volume, Leelo Keevallik, Jan Lindström,
Yael Maschler, Simona Pekarek Doehler, for their highly constructive comments and sugges-
tions on earlier versions of this chapter, which have helped improve its quality; and Elizabeth
Carter for her much appreciated editorial assistance. Standard disclaimers apply.
https://doi.org/10.1075/slsi.32.06wan
152 Wei Wang and Hongyin Tao
‘feel, think’ is considered to express either the feelings or opinions of the speaker,
as illustrated in the following two constructed examples (Lü et al. 1980).
(1) 我觉得热
wo juede re
I feel hot
‘I feel hot.’
(2) 我觉得应该去一趟
wo juede yinggai qu yi tang
I think should go one cl
‘I think (I/you/someone) should go.’
In these examples, (wo) juede serves as the matrix clause, followed by an object
complement (Thompson 2002, henceforth “complement”).
In the past few decades, a number of empirical studies have shown that
complement-taking verbs such as think and know have been largely grammati-
cized as epistemic or deontic formulas.2 Tao (1996, Chapter 8) notes the fact
that for Mandarin Chinese, which is generally considered a language typifying
‘zero anaphora’, the first person pronoun in such constructions is rarely dropped,
attesting to the fixed nature of epistemic constructions. Huang (2003) analyzes
the major complement-taking verbs, including juede, in Mandarin Chinese and
argues that conversation participants orient their subsequent turns towards the
complement clauses rather than the complement-taking matrix clause. Fang
(2000) likewise discusses several evidential and epistemic verbs including juede
in Mandarin Chinese from the perspective of grammatical de-categorization and
semantic bleaching, arguing that these verbs are in the process of shifting from
complement-taking predicates to pragmatic markers.
Endo (2010) presents a more comprehensive examination of the matrix clause
wo juede. She categorizes wo juede into two functional types – epistemic and
non-epistemic – and analyzes the epistemic uses of wo juede in relation to the
turn position, i.e., turn-initial, turn-medial, and turn-final. She notes the prosodic
features of wo juede that are common in formulaic expressions, such as tone loss
and syllable reduction, although these are not the main focus of her study.
Finally, more recent studies specifically examine the uses of wo juede in the
framework of conversation analysis. Lim (2009, 2011), for example, proposes two
interactional uses of wo juede: 1) to position the speaker’s pre-emptive awareness of
the recipient’s possible objection to a proposition; 2) to initiate a joint assessment.
2. For some sample studies, see Thompson & Mulac 1991b; Biber et al. 1999; Diessel & Tomasello
2001; Thompson 2002; Tao 2003a; Maschler 2012; and Laury & Helasvuo 2016.
Chapter 6. From matrix clause to turn expansion 153
While the previous studies have provided valuable insights into the syntactic
and discourse characteristics of wo juede, in this chapter, we will draw attention to a
functional shift of wo juede against the backdrop of clause-combining phenomena.
In particular, we investigate a further emerging phenomenon in conversational
interaction that has not been dealt with in the literature. Specifically, we report
that wo juede ‘I feel/think’ in Mandarin has further developed a conversation
organizational function, akin to what Goodwin (1996) calls attention to with the
term ‘prospective indexical’, that helps extend a current turn. After first describing
the evaluative and epistemic uses of this emerging phenomenon in its sequential
context, we will show how it is possible for a clause complex to develop further
from an (inter)subjective device to take on a turn-expansion function. Overall,
this chapter aims to elucidate this development with evidence from syntactic con-
figuration, conversational organization, prosodic marking, and frequency effects
in language use. Our study thus corroborates findings from a number of other
languages on think/feel and comparable verbs (e.g., Kärkkäinen 2003 on English
and Lindström et al. 2016 on a number of languages of Europe and other regions).
2. Data
The data used for this study come from six conversations among friends, who
are mostly young adult speakers. The conversations are transcribed according to
conversation analysis conventions (Jefferson 2004), with the modification that
each line represents one intonation unit (Du Bois et al. 1993). The duration of the
conversations ranges from fifty minutes to two hours. Detailed information about
each conversation is provided in Table 1.
Table 1. Data sources for this study

No. File name Duration Number of speakers Gender Age range
1 CCMMZM 2 hours 4 all females 27–33
2 EVANZTYC 2 hours 4 3 males and 1 female 27–32
3 SSKLM 1.5 hours 3 2 females and 1 male 23–25
4 LJWHJJ 50 minutes 2 2 females 24–27
5 LLM 50 minutes 2 2 males and 1 female 26–32
6 ZYLK 50 minutes 2 2 females 20–22
In the 8 hours of conversation, there are 269 tokens of juede. Among them, the first
person singular pronoun, wo, is by far the most frequent subject type for juede, to-
taling 226 tokens. The second and third person singular pronouns, i.e., ni ‘you’ and
ta ‘he/she’, occur with juede 21 and 22 times, respectively, as shown in Table 2. The
remaining 11 tokens have plural pronouns (i.e., women ‘we’, nimen ‘you-plural’, ta-
men ‘they’) as subjects. This distribution is consistent with Endo (2010), who finds
that the first person singular pronoun is the predominant subject type for juede,
as well as with many recent cross-linguistic studies on conversationally frequent
complement-taking predicates such as think, guess, remember, and know.3
Table 2. Distribution of juede across subject types.

Subject type Frequency Percentage
1st singular (wo) 226* 82.9%
2nd singular (ni) 21 6.7%
3rd singular (ta) 22 7.0%
Plural (women ‘we’, nimen ‘you’, tamen ‘they’) 11 3.4%
Total 269 100%
*We note that there are 36 cases (15.9% out of 226 tokens) where auxiliaries and adverbs appear between
wo and juede, such as wo hui juede ‘I would think’, wo zhende juede ‘I really think’, wo ye juede ‘I think so
too’, and wo youde shihou juede ‘I sometimes think’. These tokens are all included here.
In our data, it is clear that this clause combination has the following prototypical
syntactic configuration:
Wo ‘I’ + juede ‘feel, think’ + complement
Figure 1. Standard syntactic configuration of the complex clause
In the following sections, we will show that this construction has emerged from
an internal subjective mental state expression to an intersubjective expression and
further developed into a device for turn expansion.
3. Subjective use of wo juede
We note that as a cognitive verb (or ‘private verb’ as in Quirk et al. (1985) and
Biber (1988)), juede has the literal meaning of ‘feel/think’. It is thus not surpris-
ing that the proposition expressed in the complement clause of juede involves the
speaker’s personal or private perspective, be it mental states, personal feelings,
sensations, or opinions, with multiple dimensions often blended (cf. Englebretson
3. See, for example, Thompson & Mulac 1991a, 1991b; Scheibman 2001; Thompson 2002; Tao
2001, 2003a, 2003b; Kärkkäinen 2003, 2007; Lindström et al. 2016; Laury & Helasvuo 2016; and
Maschler 2012, 2017.
2007: 12 on the nature of stance). In some earlier studies, a distinction is made

between the epistemic and non-epistemic uses of wo juede. For instance, Endo
(2010) distinguishes personal opinion (epistemic use) from personal sensations
and feelings (non-epistemic use). Similarly, Lim (2011) identifies wo juede as both
(i) an epistemic stance marker proffering a hedged opinion and (ii) an affective
stance marker indicating a personal feeling.
In this chapter, we propose that both these uses, regardless of these different
labels, essentially convey speakers’ subjective expression of their stance towards,
viewpoint on, or feelings about the entities or propositions that they are talking
about (Thompson & Hunston 2000), as opposed to an objective statement of a
proposition. We thus consider wo juede primarily a marker of evaluation (e.g.,
Hunston & Sinclair 2000; Thompson 2002; Du Bois 2007).
In addition to, and in connection with, the evaluative stance, wo juede also
has an epistemic function, that is, downgrading or distancing the speaker’s com-
mitment to his or her evaluation. This use, as will be argued below, occurs in a
different structural configuration, with wo juede in a post-evaluation position.
The formal distinction between evaluative and epistemic stance marking can be
schematized as follows:
Evaluative stance:
Wo juede + complement clause
Epistemic stance:
Clause + wo juede
Figure 2. Formats for evaluative and epistemic uses of wo juede: The subjective domain
We discuss these two uses, both considered to be subjective, next.
3.1 Evaluative use
Wo juede can index a wide range of evaluations, from the speaker’s personal or
private domain to the public domain.
Example (3) below demonstrates wo juede being used to express the speaker’s
mental state. Jing, a female college student originally from a rural area in China, tells
her female friend Jiawei, who grew up in a big city, about her childhood experiences.
(3) Two female classmates reminiscing about childhood life in rural China
(LJWHJJ_03)
1 Jing: 我们农村的孩子很朴素的，
women nongcun de haizi hen pusu de
we village pt kids very simple pt
‘Village kids like us are very simple,’
2 就坐在一起聊天啊，
jiu zuo zai yiqi liaotian a
just sit in together chat pt
‘just sit together chatting,’
3 然后去玩玩那个什么游戏啊
ranhou qu wanwan nage shenme youxi a
and-then go play dm what game pt
‘and then play (some kind of game),’
4 Jiawei: 我突然间我好有优越感，
wo turanjian wo hao you youyuegan
I suddenly I very have sense-of-superiority
‘I suddenly have a sense of superiority,’
5 > 我觉得我在参加变形计 hh，
wo juede wo zai canjia Bianxingji
WO JUEDE I was participating X-Change
‘WO JUEDE I was participating in X-Change,’
‘It felt like I was in the X-Change game,’
6 听他们农村的人在讲大山里面的故事，
ting tamen nongcun de ren zai jiang dashan limian
listen-to them countryside pt people asp tell mountain inside
de gushi
pt story
‘listening to people in the countryside telling stories from the mountains,’
In this extract, Jing provides a description about her childhood life in a rural vil-
lage, which is totally alien to Jiawei, who was brought up in a big city with modern
electronic devices. Struck by Jing’s experience, Jiawei, in line 4, jokes that she
suddenly gets a feeling of superiority. To explicate this feeling, Jiawei uses wo juede
to introduce an evaluation about her current mental state – that she feels it is like
being in a reality show, X-Change (3. 5), where two teenagers from completely dif-
ferent backgrounds swap places and live with each other’s families for one month.
Hearing wo juede, the irrealis nature of its complement clause is successfully
interpreted by Jing.
In our data, there are 21 tokens of wo juede used for evaluations in speakers’
private or personal domain, accounting for 9% of all its occurrences. The majority,
i.e., 182 tokens (81%), are used for evaluations in the public domain. That is, the
entities or states of affairs being evaluated are not about the speakers themselves,
but about something in the external world. Extract (4) below provides such a case.
(4) Two friends talking about rifles (EVANZTYC_017)
1 ZT: 那个是不是后坐力特大，
nage shibushi houzuoli te da
that yes-no-yes recoil extremely large
‘Is the recoil (of rifles) extremely strong?’
2> Evan: 我觉得那个后坐力很大，((nodding))
wo juede nage houzuoli hen da
WO JUEDE that recoil very large
‘wo juede its recoil is very strong,’
3 而且它不是说，
erqie ta bushi shuo
in-addition it not say
‘and it’s not that,’
4 它是用手端着，
ta shi yong shuo duanzhe
it is with hand hold
‘(you have to) hold it with your hands,’
In example (4), two friends, ZT and Evan, are talking about rifles. When ZT asks
whether rifles have extremely large recoil forces (4. 1), Evan, who is more knowl-
edgeable about guns, gives an affirmative answer (4. 2). This answer is delivered in
the form of an evaluation prefaced by wo juede, where the stance object houzuoli
‘recoil’ belongs to the external world rather than the speaker’s internal world.
3.2 Epistemic use
Epistemicity is commonly understood as relating to a speaker’s knowledge and

belief: epistemic stance markers have been claimed to express one’s degree of com-
mitment to, or confidence in, what one says (e.g., Palmer 1986; Coates 1990; Bybee
et al. 1994). However, some scholars, taking an interactional approach, have begun
to shift away from this semantic definition to the interactional accomplishments
of epistemic markers, such as boosting or attenuating the illocutionary force
of speech acts (Holmes 1982, 1984) and attending to the needs of the recipient
(Kärkkäinen 2003).
In the case of wo juede, we take an inclusive perspective: semantically, a small
number of tokens do show an epistemic function that lowers the speaker’s certainty
towards his or her statement. Interactionally, wo juede reflects an orientation to the
recipient, a deliberate calibration of the relative knowledge status of conversation

participants, and an online negotiation of one’s displayed stance. Structurally, the
epistemic wo juede takes a distinctive format, where wo juede is positioned follow-
ing the evaluation. Two examples are provided below.
Extract (5) comes from a three-party conversation, where both Susie and SK
went to the same university, Peking University (PKU), while the third participant,
LM, went to a different one:
(5) Three friends talking about the popular forum on their university BBS
(SSKLM_012)
1 Susie: 北大的未名未名是:，
beida de weiming weiming shi
Peking-University pt pn pn is
‘The Weiming, Weiming (BBS) of Peking University is,’
2 Love 版，
love ban
love section
‘Love Section,’
3 > 最(.) 比较火我觉得.
zui bijiao huo wo juede
most relatively popular wo juede
‘is the most, a relatively more popular, wo juede.’
4 SK: Love 版，
love ban
love section
‘Love Section,’
5 我好久没上了.
wo haojiu mei shang le
I long-time not visit pt
‘I haven’t visited (it) for a long time.’
Prior to this sequence, LM had been talking about the most popular forum on
their university’s online bulletin board (BBS). Susie then picks up on this topic and
tells LM about her university’s most popular forum. Lines 1–3 offer Susie’s frag-
mented evaluation about Weiming BBS, with a repetition in line 1 (i.e., Weiming,
the name of the BBS) and a reformulation in line 3, where she starts with zui
‘most’ and modifies it as bijiao huo ‘relatively popular’ after a brief pause. This
reformulation downgrades the initial evaluation that ‘the Love Section is the most
popular on Weiming’ to ‘the Love Section is a relatively popular one on Weiming’.
The downgrading process is finished with wo juede, which indexes the subjective
nature of her statement and thus further lowers Susie’s epistemic commitment.
This series of interactional movements cannot be adequately understood without

a close analysis of the participation framework. Since both Susie and SK went to
the same university, they can be assumed to share equal epistemic access to the
entity being evaluated – their university BBS. Thus, Susie’s epistemic downgrading
move can be seen as oriented to her fellow PKU graduate, SK’s, epistemic status,
without sounding assertive. Additionally, the turn-final wo juede also provides
another possible place of turn completion and thus another opportunity to invite
a response from the recipient.
Extract (6) is another example of wo juede being used to lower the speaker’s
epistemic commitment. Three friends are talking about a mutual friend in rela-
tion to a tough professor; prior to this exchange, MM had been talking about
Eason, a tough professor in her department, who often fails students in their
PhD qualifying exams:
(6) (CCMMZM_020)
1 ZM: 子锦是Eason的学生啊，
zijin shi Eason de xuesheng a
pn is pn pt student pt
‘Zijin is Eason’s student?’
2 Susie: 对，
dui
right
‘Right,’
3 那(.)子锦有没有觉得岌岌可危呀，
na zijin you meiyou juede jijikewei ya
then pn have have-not feel insecure pt
‘then did Zijin feel insecure?’
4 MM: 子锦有，
zijin you
pn have
‘Zijin did,’
5 > 我觉得有.
wo juede you
wo juede have
‘wo juede (she) did.’
In line 1, ZM seeks confirmation as to whether Zijin, their mutual friend, was a

student of the tough professor Eason. Susie, who studies in the same department,
confirms (6. 2) and then turns to MM, asking whether Zijin felt insecure (6. 3).
MM, who knows Zijin better than the other two participants, first provides an
assertive answer zijin you ‘Zijin did (feel insecure)’ in line 4, which involves an
evaluation of Zijin’s possible feelings. Yet immediately in line 5, MM modifies her

response by employing wo juede, which essentially downgrades her previously
displayed stance by limiting her assertion to just her own speculation.
To sum up, the matrix clause combination wo juede is primarily used for
speakers’ subjective evaluations of entities or states of affairs related to themselves
and/or to the external world. Additionally, the non-canonical, post-positioned
wo juede functions as an epistemic marker, downgrading the speaker’s commit-
ment to their statements. Working with English data, Kärkkäinen (2003) argues
that I think, an equivalent item in English, may index a whole range from some
uncertainty to a high degree of uncertainty; the placement of the primary accent
(i.e., whether it is on think or not) suggests the specific level of epistemic stance
conveyed in the utterance. In our case, it is the position of wo juede relative to
the evaluative statement that seems to play this role: the post-positioned wo juede
emphasizes the speaker’s downgraded commitment, while the pre-positioned wo
juede displays more certainty. Note that the epistemic use of wo juede should be
viewed as a dynamic interactional process – rather than a fixed degree of epistemic
access – in which the speaker’s stance is lowered as a result of a careful calibra-
tion of all participants’ epistemic status and an online negotiation of the speaker’s
own displayed stance.
4. Intersubjective use of wo juede
While the majority of the cases express a speaker’s subjective evaluative stance, we
have also observed a number of tokens with a strong intersubjective orientation.
Broadly speaking, intersubjectivity refers to shared understanding (see Duranti
2010) or possible affinity between people’s perspectives, e.g., Schegloff (1992) and
Gillespie and Cornish (2010). This term is subject to different interpretations, how-
ever. For instance, Mori and Hayashi (2006) define intersubjectivity as agreement
on a shared definition of an object, while Heritage (2007) construes intersubjectiv-
ity as common recognitional reference, in so far as referring is concerned, which is
sometimes said to be in conflict with the progressivity of interaction. In the present
study, we approach intersubjectivity in terms of the recipient’s understanding of,
and/or affiliation with, the speaker’s stance conveyed in the prior utterance.
In terms of sequence organization, it is noted that most of the intersubjective
expressions are found in the second position (Schegloff 2007), in response to a
prior speaker’s turn. In our data, speakers are often found to employ wo juede to es-
tablish affiliation with the previous speaker by displaying a shared evaluation. This
type of evaluation with wo juede has two prominent structural characteristics: 1) it
typically has an incomplete complement or no complement at all; 2) adverbs and
referential forms tying to the prior speaker’s turn, such as the additive ye ‘also, too’,
and the demonstrative pronoun zheyang ‘like this’, are commonly used. There are
13 intersubjective cases of wo juede in the data, of which 8 tokens are accompanied
by either ye or zheyang. This use can be schematically represented as follows:
Speaker 1: Turn
Speaker 2: Wo + adverbial(s) + juede + (indexicals)
Figure 3. Intersubjective use of wo juede
In other words, there is no complement clause after the cognitive verb, with the
unexpressed evaluation inferred from the larger conversational sequence. Two
examples are provided below to illustrate this use.
(7) Three close friends commenting on homemade coffee (CCMMZM_033)
((Prior to this extract CC had been describing how she made coffee at
home.))
1 CC: 然后就自己泡咖啡，
ranhou jiu ziji pao kafei
then just self make coffee
‘Then I just make coffee myself,’
2 MM: 好喝吗?
haohe ma
tasty pt
‘Is it tasty?’
3 CC: 比外面买的咖啡好喝无数倍，
bi waimian mai de kafei haohe wushu bei
than outside buy pt coffee tasty countless times
‘A million times better than the coffee you buy elsewhere,’
4 > Susie: 我也觉得，
wo ye juede
wo also juede
‘wo also juede, /I feel the same, /I think so too,’
5 CC: 对，
Dui
Right
‘Right.’
In this excerpt, CC offers the assessment that her home-made coffee is ‘a mil-
lion times better’ than what one gets in a coffee shop (7. 3). Aligning with CC,
Susie indexes a shared stance by employing wo ye juede ‘I feel/think so too’ (7. 4).
This affiliation (Stivers 2008) is acknowledged by CC with an affirmative token

dui ‘right’ (7. 5).
In example (8), there are multiple instances of wo juede, with two intersubjec-
tive tokens occurring in lines 5 and 7. Right before this excerpt, Evan had expressed
his opinion that if one wanted to stay in good shape, s/he had to work out regularly
and eat very carefully:
(8) YC, ZT, and Susie exchanging opinions about exercising (EVANZTYC_021)
1 Susie: 我觉得,
wo juede
I think
‘I think,’
2 我可以理解，
wo keyi lijie
I can understand
‘I can understand,’
3 但是对我而言:，
danshi dui wo eryan
but for me regarding
‘but for me,’
4 还是^ too much,
haishi too much
still too much
‘(it would be) too much,’
5 > YC: 我觉得，
wo juede
wo juede
‘wo juede,’
6 ZT: hhh.
7 > YC: 我觉得也是这样我会跑步但是:，
wo juede yeshi zheyang wo hui paobu danshi
wo juede also like-this I would jog but
‘I think so too, I would jog but’
8 Susie: 对对对，
dui dui dui
right right right
‘Right, right, right,’
At the beginning of the extract, Susie, while expressing her affiliation with Evan’s
claims about exercising regularly, suggests that it would be too much for her
personally (8. 1–4). YC then shows strong affiliation with Susie’s stance: after his
turn initiation (i.e., wo juede, in 8. 5), and ZT’s laughter (8. 6), YC reinitiates his
evaluation (8. 7), which is initially framed as an alignment with Susie, since the
indexical form zheyang ‘this way/like this’ refers back to Susie’s just-prior stance.
This evaluation is immediately spelled out in the same intonation unit (8. 7). YC
ends his turn with a conjunction danshi ‘but’ without specifying his contrastive
stance as implied by the conjunction. Susie then takes over the floor with dui dui
dui ‘right, right, right’ (8. 8), suggesting affiliation with the stance taken by YC.
Notice that in line 7 (and in fact 8. 5 as well) no complements are used in this case
even when there is talk after wo juede (8. 7). The evaluation expressed here by
YC is in second position, in response to the initial opinion provided by the other
speaker, Susie. These features typify the cross-speaker, cross-turn, and responsive
nature of intersubjective uses of wo juede.
5. Wo juede and turn expansion
So far we have shown the subjective (i.e., evaluative and epistemic) and intersub-
jective uses of wo juede. These uses all involve an evaluation of some kind – some
expressed in the following object complement, others in the clause preceding wo
juede, while still others might be retrievable from the turn of the previous speaker.
However, we have found in our data a distinctive usage pattern, where wo juede
is primarily used to expand a speaker’s current turn by projecting an upcoming
multi-unit utterance.
Occurring either turn-initially or turn-medially (to be detailed below), wo
juede signals that the speaker has a point to make or a story to tell which will
take multiple turn units. In typical instances of this use, wo juede manifests itself
purely as a device for turn expansion without much of an evaluative stance. Yet
there are still other cases where some traces of evaluation can be detected: these
can be viewed as being in a transitional stage from (inter)subjective use to turn-
organizational use. Despite the residual evaluative stance, this type of wo juede can
be clearly distinguished from the (inter)subjective use, since the evaluation, if any,
does not occur immediately after wo juede, but may appear several units later. This
is illustrated by Extract (9).
(9) Three college friends discussing Li’s experience of being a fan of a singing
group (LJWHJJ_001)
1 Susie: 那你是他们的什么粉丝协会的吗?
na ni shi tamende shenme fensi xiehui de ma
then you are their what fan club pt pt
‘Are you in their fan club?’
2 Li: 我不- 没有那么脑残，((shaking head))

wo bu meiyou name naocan
I neg neg so retarded
‘I don’t-, (I’m) not so stupid,’
3 Jing: 十几年应该是那种程度吧，
shiji nian yinggai shi nazhong chengdu ba
more-than-ten years should be that-kind degree pt
‘(Being a fan for) more than ten years should be like that, shouldn’t
it?’
4 Li: 没有那么脑残啦，
meiyou name naocan la
neg so retarded pt
‘(I’m) not that stupid,’
5 我觉得那个还蛮脑残的，
wo juede nage hai man naocan de
wo juede that adv quite retarded pt
‘wo juede that is really stupid,’
6 那个，
nage
that
‘that,’
7 > 我觉得，
wo juede
wo juede
‘wo juede,’
8 就是怎么讲，
jiushi zenme jiang
adv how to-say
‘how to put it,’
9 没有那么夸张啦，
meiyou name kuazhang la
neg so crazy pt
‘(I’m) not that crazy,’
10 我很受不了就是那种就是跟着，
wo hen shoubuliao jiushi nazhong jiushi genzhe
I very cannot-stand adv that-kind adv follow
‘I really cannot stand that kind (of fan who) follows (their idol) around,’
11 ^我爱你我爱你::， ((mimicking those crazy fans))
wo ai ni wo ai ni
I love you I love you
‘(proclaiming) I love you I love you!’
Prior to extract (9), Li has talked about her experience as an enthusiastic fan of a
singing group. This triggers Susie’s question, whether Li was ever in their fan club
(9. 1). After Li’s denial, Jing expresses a doubt about Li’s denial (9. 3), speculating
that Li must have been a die-hard fan given the length of her fanhood. This gives
rise to Li’s extended clarification (9. 4–11). Li first repeats her denial (9. 4), and
then provides an evaluation about those crazy fans, distancing herself from them.
The turn-expansion wo juede occurs in line 7 to project a multi-unit utterance, in
which she first emphasizes the fact that she is not as crazy as typical fans (9. 9),
and then expresses her attitude toward the behavior of some fans (9. 10–11). One
might argue that line 10 does indeed involve some kind of an evaluation. While
that may be the case, what should also be noted is the distance and the amount of
materials appearing between it and wo juede: a hedging expression (i.e., ‘how to
put it’ in 9. 8) and a repetition of denial (9. 9). The fact that wo juede is produced
in an independent intonation unit (more on this in sect. 5.1) and not immediately
followed by any evaluation leaves it open for the speaker to expand her turn.
This use of wo juede can be considered least prototypical and thus ‘newly
emerging’ for several reasons: (1) it has little to do with the lexical semantics of
wo juede, namely as a formula expressing the mental activity or private feeling of
a person; (2) there is no indication of evaluation that can be directly and/or im-
mediately related to wo juede; and (3) it is only observed in the spoken language,
where linguistic innovations take place first (Chafe 1982).
An additional example is provided below to illustrate this use. In extract (10),
CC is completing a recount of her fieldwork in a rural village in China (10. 1–6).
Her main point, as shown in prior utterances, is the fact that people in the coun-
tryside tend to argue over trifling matters:
(10) On fieldwork and rural life (CCMMZM_002)
1 CC: 如果你单是去那边问他们二十分钟，
ruguo ni danshi qu nabian wen tamen ershi fenzhong
if you just go there ask them twenty minutes
‘If you just go there and ask them (questions) for twenty minutes,’
2 没有办法问一些事情.
meiyou banfa wen yixie shiqing
no way ask some thing
‘(you have) no way to know some things (that you are interested in),’
3 对.
dui
right
‘Right.’
4 (2.0)
5 就干这些事情，
jiu gan zhexie shiqing
adv do these things
‘(That’s) what I did,’
6 超八卦的.
chao bagua de
super gossipy pt
‘super gossipy.’
7 MM: hhhh.
8 > MM: 而且我觉得像农村，
erqie wo juede xiang nongcun
in-addition wo juede like countryside
‘In addition, wo juede, like the countryside,’
9 就像我奶奶家就是农村里，
jiu xiang wo nainai jia jiu shi nongcun li
just like my grandmother home adv is countryside in
‘just like my grandma’s home is in the countryside,’
10 就是，
jiushi
dm
‘well,’
11 他们(.)关注的事情真的就是特别特别小的一件事情,
tamen guanzhu de shiqing zhende jiushi tebie tebie xiao
they care pt things really are extremely extremely small
de yi jian shiqing
pt one CL thing
‘what they care about are those extremely trivial things,’
12 CC: 嗯.
en
yeah
‘Yeah.’
13 MM: 就比如说，
jiu birushuo
just for-example
‘For example,’
14 你家把我家的树苗怎么[了，
ni jia ba wo jia de shumiao zenme le
your house pt my house pt sapling what pt
‘what did your folks do to our saplings,’
15 CC: [对对对，
dui dui dui
right right right
‘Right, right, right,’
After CC’s turn reaches a completion point in line 6 and ZM laughs (10. 7), MM
launches a multi-unit turn projected by wo juede (10. 8), in which she brings in
her own experience with rural life. Wo juede signals that the speaker has a point to
make in the upcoming turn, which will be delivered in multiple units. As can be
seen in lines 8–11, MM first establishes the topic ‘countryside’, and then explains
her source of knowledge, that is, her grandma lives in the countryside. Finally,
in line 11, she delivers the main point, that people in the countryside care about
very trivial things. The early projection of the multi-unit utterance affords the
speaker the opportunity to develop a point to the desired degree without: 1) being
constrained by the regular turn space; and 2) the risk of being cut off by another
conversation participant.
Having discussed the interactional characteristics of emerging wo juede, we
now turn to its prosodic design. We have found two types of turn-expanding wo
juede, which differ in their intonational status – wo juede as an independent into-
nation unit and as part of a larger intonation unit. These two types have nuanced
interactional import and will be discussed below (for more details on prosodic
features, see Wang 2017).
5.1 Prosodically integrated wo juede
As a turn-expanding device, wo juede can be embedded in a larger intonation

unit. Within the same intonation unit, what immediately follows wo juede is an
NP or a clause that serves as the topic to be developed. Thus, wo juede can be seen
as establishing the topic for the projected multi-unit utterance. This use can be
schematized as Figure 4 below:
Speaker 1: Turn.
Speaker 2: (Clauses,)
wo juede + NP/clause,
Figure 4. Format of prosodically integrated wo juede
Example (10) above is a case of prosodic integration. Below we provide another

example with a pitch contour graph to illustrate its prosodic integration.
(11) Two college classmates talking about a Japanese class and its instructor
(ZYLK_058)
((Prior to this extract, Kai told Yi that her Japanese instructor gave them
extremely dry and technical lectures and she felt disoriented about the exam.))
1 Kai: 我说考试不会考这种吧，
wo shuo kaoshi bu hui kao zhe zhong ba
I say exam not will test this kind pt
‘I said this kind (of grammar analysis) will not be tested in the exam?’
2 我说完全记不住啊，
wo shuo wanquan jibuzhu a
I say completely cannot-remember pt
‘I said I will remember nothing,’
3 Yi: 应该会考，
yinggai hui kao
should will test
‘It will be tested for sure,’
4 那你死定了，
na ni si ding le
then you die sure pt
‘then you are doomed for sure,’
5 Kai: 死定了的感觉，
si ding le de ganjue
die sure pt pt feeling
‘(I have a) feeling that I’m screwed,’
6 (1.8)
7 > 而且我觉得她，
erqie wo juede ta
additionally wo juede she
‘additionally wo juede she,’
8 老师还说，
laoshi hai shuo
teacher also say
‘the teacher also said,’
9 没事儿你们这一页的那个:，
meishi’er nimen zhe yi ye de nage
never-mind you this one page pt dm
‘never mind, you guys, this page, well,’
10 就是标出来那些:大写字母不需要背啊，
jiushi biao chulai naxie daxie zimu bu xuyao bei a
dm mark comp those capital letter not need memorize pt
‘those marked in capital letters don’t need to be memorized,’
11 我们说不背你干嘛在黑板上写得这么顺，
women shuo bu bei ni ganma zai heiban shang xie de
we say not memorize you why on blackboard loc write pt
zheme shun
this smooth
‘we said/thought, (if they) don’t (need to be) memorized, why did you
write them out on the blackboard to begin with?’
At the beginning of example (11), Kai expresses her concern about her Japanese
examination, as the instructor teaches grammar in a very technical way, which
is often beyond her comprehension. Yi thus jokes that Kai is doomed to fail the
exam (11. 3–4), and Kai aligns with Yi (11. 5). After a 1.8-second pause, which
suggests the completion of that sequence, Kai chooses to continue her turn with
additional details about her Japanese class. In line 7, she uses wo juede to project
a forthcoming multi-unit utterance on the one hand, and to introduce the topic
of her following talk on the other. The topic being introduced is initially ta ‘she’,
which might not be identifiable for Yi. Thus, after a brief pause, she switches to a
more specific reference term laoshi ‘teacher’.
The prosodic features of this token, analyzed with Praat (Boersma & Weenink
2018), are shown in Figure 5. Wo juede is embedded in the intonation unit erqie wo
juede ta ‘additionally WO JUEDE she’; a clear gap is observed after the topic ta ‘she’.
0 1.62258503
500
400
300
Pitch (HZ)
200
75
erqie wo juede ta laoshi hai shuo
additionally WO JUEDE she teacher also said
0 1.623
Time (s)
Figure 5. Prosodically integrated wo juede: ‘Additionally WO JUEDE she, the teacher also
said’
5.2 Prosodically independent wo juede
Alternatively, as alluded to earlier in the analysis of extract (9), wo juede can take
an independent prosodic form, occupying a whole intonation unit separated from
neighboring units by a brief pause. Interactionally, this type of wo juede is simply
a projecting device without the topic-introducing function. Usually the indepen-
dent wo juede is prosodically more prominent, with longer duration, larger pitch
range, and without any tonal reduction to neutral tone (Wang 2017).
This use is schematized as follows (Figure 6):
Speaker 1: Turn.
Speaker 2: (Clauses,)
wo juede,
Figure 6. Format of prosodically independent wo juede
In example (12), wo juede is produced in response (12. 7) to Yi’s informing that she
cares about breakfast, unlike her roommate, who cares more about dressing up. Wo
juede again works to project that there will be a multi-unit utterance forthcoming.
(12) Yi and Kai on the dilemma of breakfast or more sleep (ZYLK_64)
1 Yi: 我室友在那边很着急地化妆穿衣服，
wo shiyou zai nabian hen zhaojide huazhang chuan yifu
my roommate in there very hurriedly make-up put-on clothes
‘My roommate is there putting on make-up and clothes hurriedly,’
2 我就在那边很着急地吃早餐，
wo jiu zai nabian hen zhaoji de chi zaocan
I adv in there very hurriedly pt eat breakfast
‘I’m there eating breakfast hurriedly,’
3 hh,
4 Kai: hh.
5 Yi: 大家迟到hh，
dajia chidao
everybody late
‘(As a result,) everybody is late.’
6 Kai: 哎哟我一般-,
aiyo wo yiban
exl I usually
‘Well, I usually,’
7 > 我觉得,
wo juede
wo juede
‘wo juede,’
8 像:一周有两天起早床我就吃早餐，
xiang yizhou you liang tian qizaochuang wo jiu chi zaocan
like a-week have two day get-up-early I adv eat breakfast
‘like, there are two days in a week when I (need to) get up early, then
I eat breakfast,’
9 然后剩下三天就睡了h，
ranhou shengxia san tian jiu shui le
and-then remaining three days adv sleep pt
‘and then, I sleep in the remaining three days,’
10 就直接起来吃午餐，
jiu zhijie qilai chi wucan
just directly get-up eat lunch
‘then I get up and go straight to lunch,’
11 Yi: 其实我不知道为什么你们那么能睡哎，
qishi wo bu zhidao weishenme nimen name neng shui ai
honestly I neg know why you so be-able sleep pt
‘Honestly I don’t know how you guys can sleep so much,’
12 觉得很，
juede hen
feel very
‘(I) feel (it’s) very,’
13 Kai: 我我感觉:可能不需要吃很多东西，
wo wo ganjue keneng bu xuyao chi henduo dongxi
I I feel probably neg need eat much food
‘I, I feel, (I) probably do not need to eat much food,’
14 但是一定要睡得饱饱的，
danshi yiding yao shui de baobaode
but for-sure have-to sleep pt enough
‘but have to sleep enough for sure,’
As shown in Figure 7, the token at line 7 is intonationally independent, with no-

ticeable pauses before and after, separating it from neighboring units. Also, its pro-
sodic prominence is manifested by long duration (312 ms) and a large pitch range
(46 Hz), in contrast to the token in (9) produced by the same speaker – 302 ms and
20 Hz. Finally, all three syllables are fully pronounced without any tonal reduction.
3.58609084 4.05707112
400
300
200
Pitch (HZ)
70
aiyo wo yiban wo juede xiang
EXL I usually WO JUEDE like
1.888 4.196
Time (s)
Figure 7. Prosodically independent wo juede: “I usually, WO JUEDE, like”
Interactionally, this token does not indicate any sort of evaluative stance, since
what follows it is not an evaluation but an informing about a fact concerning
herself, that is, that she eats breakfast twice a week when she has to get up early.
Wo juede essentially expands the current speaker’s turn by adding new materials.
Its prosodic prominence and independence match the loose relationship between
wo juede and the added material in the expanded turn, effectively reinforcing the
status of wo juede as a turn-organizational device.
Goodwin (1996) points out that in conversation speakers sometimes provide
a linguistic expression whose specification is to be discovered in subsequent talk.
Such expressions, such as “a problem” in a statement as “we definitely have a
problem here”, or “funny thing” in a story-preface as “a funny thing happened to
me today”, often project more to come and both parties orient to the multi-unit
process that follows. He calls these expressions ‘prospective indexicals’ (Goodwin
1996: 384).4 In the case of wo juede, we can say that its turn expansion function is
akin to a prospective indexical as it projects more to come on a particular topic.
To sum up, wo juede can serve as a turn-expansion device, projecting an
upcoming multi-unit utterance. It can be used turn-initially in response to the
prior speaker’s turn, which illustrates or strengthens the prior speaker’s point. Wo
4. For a similar point but on a different construction known as “extraposed assessing phrase”
(e.g., ‘it’s funny’), see Couper-Kuhlen and Thompson (2008).
juede can also be used turn-medially, projecting an extended utterance serving

to develop one’s own point. Among the 10 tokens found in our data, two differ-
ent prosodic designs have been identified with nuanced interactional import: 1)
intonationally integrated wo juede functions to establish the to-be-discussed topic;
and 2) intonationally independent wo juede, which is also prosodically stronger
than the integrated type, serves as a purely turn-projecting device.
6. Discussion
6.1 Functional distribution
Our analysis so far shows that wo juede is prototypically a clause combination

which has evolved into an evaluative marker – with either a subjective or an inter-
subjective orientation – and an epistemic marker lowering a speaker’s epistemic
commitment. It has also developed into a conversation organizational device, akin
to a prospective indexical in the sense of Goodwin (1996), to project the expansion
of a turn. For comparison, the frequency of each function is provided in Table 3.
Table 3. Frequency and distribution of the four major functions of wo juede

Function Frequency Percentage
Subjective: evaluative stance 198 88%
Subjective: epistemic stance 5 2%
Intersubjective: affiliative evaluation 13 6%
Turn expansion 10 4%
Total 226 100%
In terms of syntactic-conversational features, we see two polar directions of devel-

opment, one without an expressed complement ((inter)subjective), and the other
with an extended turn (turn expansion): In both of the evolved cases (indicated in
boldface in Figure 8), the scope of the extension goes beyond the standard “matrix
clause + complement clause” configuration, as they cut across multiple structural
units and/or multiple speakers.
–no complement
Wo + juede + –complement
–multi-unit extended turn
Figure 8. Polar developments in the “matrix clause” construction

6.2 Explaining the link between multiple functions
Given these diverse functions as well as the morphosyntactic configurations

observed in the data, a question arises as to the inherent link and mechanism
of development among the multiple functions and forms. Our proposal is that
expressions such as wo juede ‘I feel/think’ have special qualities that enable them
to serve interactional purposes in unique ways, fostering further development.
As stated at the beginning of the chapter, juede ‘feel’ indicates a private feel-
ing or sensation of the speaker. This feature implies that the state of affairs or
the evaluation expressed is the speaker’s personal perspective; with wo juede the
speaker thus makes no claim as to whether other speakers will agree. This char-
acteristic leads naturally to wo juede being used in situations where the speakers
deems it desirable to make explicit that this is their own perspective and to limit
the generalizability of their claim or evaluation.
6.3 Pathways of the emergence of the extended uses
While the preceding section may help explain the underlying link among the vari-
ous uses, we are equally interested here to understand why the complement form
may get lost and the erstwhile matrix clause is used as a turn-expansion device for
turns-at-talk, mostly devoid of overt evaluations. What brings about this kind of
grammatical reorganization?
Our explanations concern interactional and usage-based factors.
Interactionally, we propose that evaluations, no matter how personal or private
they are, are subject to reformulation, expansion, and negotiation between partici-
pants (Pomerantz 1984; Goodwin & Goodwin 1992). As the process gets expanded,
the connection between the matrix clause and the complement (evaluation) can
become loose to the extent that the erstwhile syntactic complement is weakened or
lost altogether. This paves the way for the matrix clause to be used with no comple-
ment clause attached to it, and to develop further into a turn-expansion device.
Example (13) shows a gradually expanded and carefully reformulated evalua-
tion produced in multiple intonation units:
(13) Susie joking about an exchange student from the U.S. (SSKLM_014)
1 > Susie: 我觉得他是，
wo juede ta shi
wo juede he is
‘wo juede he is,’
2 他是那个，
ta shi nage
he is that
‘he is,’
3 他是，
ta shi
he is
‘he is,’
4 卧底，
Wodi
Undercover
‘undercover,’
5 奸细.
Jianxi
(despicable)spy
‘spy.’
Prior to this extract, another participant had told the story of an exchange student
who studied at Tsinghua University in China, and commented that Tsinghua was
the best university in China. Susie, as a graduate of Tsinghua’s rival university,
did not believe that this person had ever studied there because of his untruthful
remarks about it. In line 1, she uses wo juede to project an upcoming evaluation,
followed by the person being evaluated, ta ‘he’. In line 2, Susie employs a pronoun,
nage ‘that’, as a placeholder for a forthcoming noun. After a partial repetition
in line 3 (i.e., ta shi ‘he is’), the evaluative term, wodi ‘undercover’, is eventually
produced (13. 4). Yet the evaluation is not yet closed. In line 5, Susie utters the
second, negative evaluative term, jianxi ‘(despicable) spy’, which often goes with
wodi ‘undercover’, but with a stronger negative connotation. The repetition, hesita-
tion, and negative semantics in Susie’s production of the evaluation manifest her
online planning process. This process inevitably increases the distance between
wo juede and the actual evaluation segment, thus weakening the evaluative force
carried by wo juede.
The next example, (14), where two speakers comment on the cuisine of
Chongqing, demonstrates how the interaction between two speakers contributes to
the lengthened assessment process, thus weakening the evaluative force associated
with wo juede. In this case two interlocutors negotiate and collaborate in the forms
of overlapping speech and collaborative finish at 14. 2–3, which resembles what
Lim (2009, 2011) describes as a joint-assessment enabling function of wo juede.
(14) Kai and Yi discussing Chongqing cuisine (ZYLK_034)

1 Kai: 所以说(.)我觉得重庆那边的人就是把辣椒做出了一种:，
suoyi shuo wo juede Chongqing nabian de ren jiushi ba
so say wo juede pn that-place pt people dm pt
lajiao zuochu le yizhong
chili make pt a-kind-of
‘So wo juede people in Chongqing have made chili peppers a kind of,’
2 [一种，
Yizhong
a-kind-of
‘a kind of,’
3 > Yi: [嗯自己的风味，
en ziji de fengwei
yeah self pt local-style
‘Yeah their own style,’
4 Kai: 对，
Dui
Right
‘Right,’
5 special 的感觉，
special de ganjue
special pt feeling
‘special feeling,’
6 Yi: 对,
Dui
Right
‘Right,’
Kai first projects an evaluation using wo juede in line 1. While searching for an
appropriate word to describe the creative way people in Chongqing cook with chili
peppers, Kai lengthens yizhong ‘a kind of ’ at the end of line 1 and then repeats this
word in line 2. Picking up the word search indicators (e.g., Lerner 1996; Hayashi
1999), Yi volunteers a candidate evaluation, thus collaboratively completing the
evaluative action (14. 3). Yi’s candidate completion is explicitly accepted by Kai, as
evidenced by her agreement token, dui ‘right’.
In both extracts (13) and (14), then, the evaluation sequence is much extended
and the link between the matrix clause (in 13. 1 and 14. 1) and the complement
clauses, if there is one, is distributed across multiple utterances. We contend that
it is through these kinds of negotiation processes that the matrix clause gradually
drifts to an independent status and exhibits greater disassociation from prototypi-
cal evaluations and their associated clause combining patterns.
An additional factor we find highly relevant is the frequency effect of language

use. Specifically, high frequency items have been shown to tend to undergo special
reduction (Bybee & Hopper 2001: 1). For example, Mulder and Thompson (2008)
show that the English conjunction but can be used “hanging” in a turn-final
position without further talk, where interlocutors have no problem treating the
turn as complete. As a result, they consider but a fully developed final particle in
(Australian) English marking contrastive content. Tao (2007) describes how the
erstwhile adverb absolutely can be used in conversational English as an indepen-
dent confirmative response token, without any otherwise expected adjectival or
verbal predicates expressed. We argue that the prospective-indexical use of wo
juede can be understood in a similar light. That is, due to the pervasive use of wo
juede as an evaluation-introducing device and in a complement-taking configura-
tion, this formula can easily come to project more to come, whereby the primary
speaker can deploy it to extend the turn while the co-participant can expect more
to come, with or without follow-throughs.5 In fact it is entirely possible that both
factors – the frequency of negotiation with evaluations and the frequency effect of
evaluations on projectability – may work together in facilitating wo juede’s becom-
ing a turn expansion device that emerges from interaction.
7. Conclusions
In this chapter we have analyzed the combination of a first person singular pro-
noun wo and a private/cognitive verb juede ‘I feel/think’ in Mandarin. We have
shown that there are a number of functions associated with this common com-
bination. What is notable is that this highly fixed construction has developed not
only evaluative and epistemic uses (in the subjective and intersubjective domains)
but further into a prospective indexical device used to extend conversational
turns, a pattern which has been reported for other languages (Kärkkäinen 2003;
Lindström et al. 2016), but not yet received much attention in the Chinese in-
teractional linguistic literature. After detailing the evaluative and epistemic uses
and the emerging turn expansion functions in Mandarin, we argued that there
are some inherent properties that enable the construction to be used in specific
interactional contexts. The result is that while a prototypical epistemic expression
involves a complex clause structure as shown in Figure 1, emerging patterns have
5. For some highly relevant yet slightly different angles on discourse markers and their
projecting/projective functions, see Pekarek Doehler (2011) on the verb ‘mean’ in French;
Maschler (2012) and Laury and Helasvuo (2016, p. 89) on the verb “know” in Hebrew and
Finish, respectively.
developed in two polar directions: one without a complement but with implied
evaluation, and the other with complex turns but without explicit or implied
evaluation, as shown in Figure 8.
We further show that the turn expansion function contrasts with the evaluative
and epistemic marking functions chiefly with regard to intonation patterns, in that
while the evaluative and epistemic uses tend to be phonologically weak all around,
the turn expansion device receives more intonational prominence (mainly by
being intonationally independent). Finally, we have also attempted to provide an
explanation for the robust historical development in the [matrix + complement]
complex, pointing to the interactive nature of the evaluative stance and the effect
of high frequency usage on projectability.
We close this chapter with three conclusions:
1. Complex clauses can be realized as a combination of a matrix clause and a
complement clause. Yet such clause combinations, under certain conditions
– and sometimes only in certain conditions (Maschler 2012) – are prone
to further grammaticization. By “certain conditions”, we mean a configura-
tion of particular subject forms and verb types for particular conversational
actions (Thompson & Mulac 1991a, 1991b; Bybee & Scheibman 1999;
Scheibman 2000, 2001).
2. Pragmatic/interactional developments involving clause combining may not be
adequately described without appealing to the larger conversation sequential
context. Traditionally, discussions of clause combining as realized in matrix
clause and complement clause connections are often limited to clause-based
syntactic structures (such as argument structure, see further discussions in
Thompson 2002). Recent studies have noted the relationship between stance
taking and expressions such as wo juede, as well as the differences in com-
municative weight between the matrix clause and the complement clause
(Thompson & Mulac 1991a, 1991b; Thompson 2002; Kärkkäinen 2003, 2012;
Huang 2003). Here we would like to echo many other researchers in their
emphasis on considering the larger sequential context within which some ma-
trix clause phenomena operate. Our Mandarin data thus provide supporting
evidence from a language that is typologically different from the oft-discussed
Indo-European languages.
3. Finally, conversational actions can be indexed with different syntactic and
prosodic patterning, which attests to the multimodal nature of language use
in talk-in-interaction. Our analysis of Mandarin conversational data shows
that the relative positioning of the stance marker and the associated evalu-
ation expression, as well as prosodic patterning, are routinely deployed for
interactional purposes – corroborating previous studies in this tradition (see,
e.g., Bybee 2001; Bybee & Scheibman 1999; Scheibman 2000; Tao 2003a; L.
Tao 2006; Maschler 2012).
Glossing abbreviations
adv adverb inj interjection

asp aspect marker neg negative morpheme
cl classifier loc localizer
comp complement pn proper noun
dm discourse marker pt particle
exl exclamation
References
Biber, D. (1988). Variation across speech and writing. Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511621024
Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). Longman grammar of
spoken and written English. London: Longman.
Boersma, P. & Weenink, D. (2018). Praat: Doing phonetics by computer [Computer program].
Version 6.0.43, retrieved 8 September, 2018 from http://www.praat.org/.
Bybee, J. (2001). Phonology and language use. Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511612886
Bybee, J., and Hopper, J. (2001). Frequency and the emergence of linguistic structure. Amster-
dam/Philadelphia: John Benjamins. https://doi.org/10.1075/tsl.45
Bybee, J., Perkins, R., & Pagliuca, W. (1994). The Evolution of grammar: Tense,aspect and modal-
ity in the languages of the world. Chicago: University of Chicago Press.
Bybee, J., & Scheibman, J. (1999). The effect of usage on degrees of constituency: The reduction
of Don’t in English. Linguistics 37 (4): 575–596. https://doi.org/10.1515/ling.37.4.575
Chafe, W. (1982). Integration and involvement in speaking, writing, and oral literature. In D.
Tannen (Ed.), Spoken and written language: Exploring orality and literacy (pp. 35–54).
Norwood, NJ: Ablex.
Coates, J. (1990). Modal meaning: The semantics – pragmatics interface. Journal of Semantics 7,
53–63. https://doi.org/10.1093/jos/7.1.53
Couper-Kuhlen, E. & Thompson, S. A. (2008). On assessing situations and events in conversa-
tion: Extraposition and its relatives. Discourse Studies 10(4): 443–467.
https://doi.org/10.1177/1461445608091882
Diessel, H., & Tomasello, M. (2001). The acquisition of finite complements clauses in English: A
corpus-based analysis. Cognitive Linguistics 12, 97–141. https://doi.org/10.1515/cogl.12.2.97
Du Bois, J. W. (2007). The stance triangle. In R. Englebretson (Ed.), Stancetaking in discourse:
Subjectivity, evaluation, interaction (pp. 139–182). Amsterdam/Philadelphia: John Ben-
jamins. https://doi.org/10.1075/pbns.164.07du
Du Bois, J. W., Schuetze-Coburn, S., Cumming, S., & Paolino, D. (1993). Outline of discourse
transcription. In J. A. Edwards & M. D. Lampert (Eds.), Talking data: Transcription and
coding in discourse research (pp. 45–89). New York/London: Psychology Press.
Duranti, A. (2010). Husserl, intersubjectivity and anthropology. Anthropological Theory 10
(1–2), 16–35. https://doi.org/10.1177/1463499610370517
Endo, T. K. (2010). Expressing stance in Mandarin conversation: Epistemic and non-epistemic
uses of wo juede. Doctoral dissertation, University of California, Los Angeles.
Englebretson, Robert (Ed.). 2007. Stancetaking in Discourse: Subjectivity, Evaluation, Interaction.
Amsterdam/Philadelphia: John Benjamins.
Fang, M. (2000). Ziran kouyu zhong ruohua lianci de huayu biaoji gongneng. [Discourse func-
tions of bleached conjunctions in natural Mandarin conversation]. Zhongguo Yuwen [Stud-
ies of the Chinese Language], 5, 459–470.
Gillespie, A., & Cornish, F. (2010). Intersubjectivity: Towards a dialogical analysis. Journal for
the Theory of Social Behaviour 40 (1), 19–46.
https://doi.org/10.1111/j.1468-5914.2009.00419.x
Goodwin, C. (1996). Transparent vision. In E. Ochs, E. A. Schegloff, & S. A. Thompson (Eds.),
Interaction and grammar (pp. 370–404). Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511620874.008
Goodwin, C. & Goodwin, M. H. (1992). Assessments and the construction of context. In C.
Goodwin and A. Duranti (Eds.), Rethinking context: Language as an interactive phenom-
enon (pp. 147–190). Cambridge: Cambridge University Press.
Hayashi, M. (1999). Where grammar and interaction meet: A study of co-participant comple-
tion in Japanese conversation. Human Studies 22 (2–4), 475–499.
https://doi.org/10.1023/A:1005492027060
Heritage, J. (2007). Intersubjectivity and progressivity in person (and place) reference. In N. J.
Enfield & T. Stivers (Eds.), Person reference in interaction: Linguistic, cultural and social
perspectives (pp. 255–280). Cambridge: Cambridge University Press.
Holmes, J. (1982). Expressing doubt and certainty in English. RELC Journal 13 (2), 9–28.
https://doi.org/10.1177/003368828201300202
Holmes, J. (1984). Modifying illocutionary force. Journal of Pragmatics 8, 345–365.
https://doi.org/10.1016/0378-2166(84)90028-6
Huang, S. (2003). Doubts about complementation: A functionalist analysis. Language and
Linguistics 4 (2), 429–455.
Hunston, S., & Sinclair, J. (2000). A local grammar of evaluation. In S. Hunston & G. Thompson
(Eds.), Evaluation in text: Authorial stance and the construction of discourse (pp. 74–101).
Oxford: Oxford University Press.
Jefferson, G. (2004). Glossary of transcript symbols with an introduction. In G. Lerner
(Ed.), Conversation analysis: Studies from the first generation (pp. 13–34). Amsterdam/
Philadephia: John Benjamins. https://doi.org/10.1075/pbns.125.02jef
Kärkkäinen, E. (2003). Epistemic stance in English conversation: A description of its interactional
functions, with a focus on I think. Amsterdam/Philadelphia: John Benjamins.
Kärkkäinen, E. (2007). The Role of I guess in conversational stancetaking. In R. Englebretson
(Ed.), Stancetaking in discourse: Subjectivity, evaluation, interaction (pp. 183–220). Amster-
dam/Philadelphia: John Benjamins. https://doi.org/10.1075/pbns.164.08kar
Kärkkäinen, E. (2012). I thought it was very interesting: Conversational formats for taking a
stance. Journal of Pragmatics 44 (15), 2194–2210.
Laury, R. H. & Helasvuo, M-L., 2016. Disclaiming epistemic access with ‘know’ and ‘remember’
in Finnish. Journal of Pragmatics 106, 80–96. https://doi.org/10.1016/j.pragma.2016.07.005
Lerner, G. H. (1996). On the ‘semi-permeable’ character of grammatical units in conversation:
Conditional entry into the turn space of another speaker. In E. Ochs, E. A. Schegloff &
S. A. Thompson (Eds.), Interaction and grammar (pp. 238–276). Cambridge: Cambridge
University Press. https://doi.org/10.1017/CBO9780511620874.005
Lim, N. E. (2009). Stance-taking with wo juede in conversational Chinese. In Proceedings of the
21st North American Conference on Chinese Linguistics, (pp. 323–340). Columbus: Ohio
State University online publications.
Lim, N. E. (2011). From subjectivity to intersubjectivity: Epistemic marker wo juede in Chinese.
In Y. Xiao, L. Tao & H. L. Soh (Eds.), Current issues in Chinese linguistics (pp. 265–300).
Cambridge: Cambridge Scholar Press.
Lindström, J., Maschler, Y., & Pekarek Doehler, S. (2016). A cross-linguistic perspective on
grammar and negative epistemics in talk-in-interaction. Journal of Pragmatics 106, 72–79.
Lü, S., Li, L., Liu, J., Fan, J., Shi, Y., Fan, F., Meng, C., Ma, S., Li, Z., Chen, J., Zhan, K., Zheng,
H., & Tao, B. (Eds.). (1980). Xiandai hanyu babaici [Modern Chinese 800 words]. Beijing:
Shangwu Yinshuguan.
Studies in Language 36(4): 785–847. https://doi.org/10.1075/sl.36.4.03mas
Maschler, Y. (2017). The emergence of Hebrew loydea/loydat (‘I dunno masc/fem’) from interac-
tion: Blurring the boundaries between discourse marker, pragmatic marker, and modal
particle. In A. Sansò and C. Fedriani (Eds.), Pragmatic markers, discourse markers and
modal particles: New perspectives (pp. 37–69). Amsterdam/Philadelphia: John Benjamins.
Mori, J., & Hayashi, M. (2006). The achievement of intersubjectivity through embodied
completions: A study of interactions between first and second language speakers. Applied
Linguistics 27 (2), 195–219. https://doi.org/10.1093/applin/aml014
Mulder, J. & Thompson, S. A. (2008). The grammaticization of but as a final particle in English
conversation. In R. Laury (Ed.), Crosslinguistic Studies of Clause Combining: The Multifunc-
tionality of conjunctions (pp. 179–204). Amsterdam/Philadelphia: John Benjamins.
https://doi.org/10.1075/tsl.80.09mul
Palmer, F. R. (1986). Mood and modality. Cambridge: Cambridge University Press.
structions in French talk-in-interaction. In R. Laury & R. Suzuki (Eds.), Subordination in
conversation: A cross-linguistic perspective (pp. 103–148). Amsterdam/Philadelphia: John
Benjamins. https://doi.org/10.1075/slsi.24.06doe
Pomerantz, A. (1984). Agreeing and disagreeing with assessments: Some features of preferred/
dispreferred turn shapes. In M. Atkinson & J. Heritage (Eds.), Structures of social interac-
tion: Studies in conversation analysis (pp. 57–101). Cambridge: Cambridge University Press.
Quirk, R., Greenbaum, S., Leech, G., Svartvik, J., & Crystal, D. (1985). A comprehensive grammar
of the English language. London/New York: Longman.
Schegloff, E. A. (1992). Repair after next turn: The last structurally provided defense of intersub-
jectivity in conversation. American Journal of Sociology 97 (5), 1295–1345.
https://doi.org/10.1086/229903
Schegloff, E. A. (2007). Sequence organization in interaction: A primer in conversation analysis.
Scheibman, J. (2000). I dunno: A usage-based account of the phonological reduction of don’t in
American English conversation. Journal of Pragmatics 32 (1), 105–124.
https://doi.org/10.1016/S0378-2166(99)00032-6
Scheibman, J. (2001). Local patterns of subjectivity in person and verb type in American English
conversation. In J. L. Bybee & P. Hopper (Eds.), Frequency and the emergence of linguistic
structure (pp. 61–90). Amsterdam/Philadelphia: John Benjamins.
https://doi.org/10.1075/tsl.45.04sch
Stivers, T. (2008). Stance, alignment, and affiliation during story telling: When nodding is a
token of preliminary affiliation. Research on Language in Social Interaction 41 (1):31–57.
https://doi.org/10.1080/08351810701691123
Tao, H. (1996). Units in Mandarin conversation: Prosody, discourse, and grammar. Amsterdam/
Philadelphia: John Benjamins. https://doi.org/10.1075/sidag.5
Tao, H. (2001). Discovering the usual with corpora: The case of remember. In R. Simpson & J.
Swales (Eds.), Corpus linguistics in North America: Selections from the 1999 symposium (pp.
116–144). Ann Arbor: University of Michigan Press.
Tao, H. (2003a). Phonological, grammatical, and discourse evidence for the emergence of zhidao
[know] constructions. [in Chinese] Chinese Language 4, 291–302.
Tao, H. (2003b.) A usage-based approach to argument structure: Remember and forget in spoken
English. International Journal of Corpus Linguistics 8, 75–95.
https://doi.org/10.1075/ijcl.8.1.04tao
Tao, H. (2007). A corpus-based investigation of absolutely and related phenomena in spoken
American English. Journal of English Linguistics 35 (1), 1–25.
https://doi.org/10.1177/0075424206296615
Tao, L. (2006). Classifier loss and frozen tone in spoken Beijing Mandarin: The yi+ge phonosyn-
tactic conspiracy. Linguistics 44, 91–133. https://doi.org/10.1515/LING.2006.004
Thompson, G. & Hunston, S. (2000). Evaluation: An introduction. In S. Hunston & G. Thomp-
son (Eds.), Evaluation in text: Authorial stance and the construction of discourse (pp. 1–27).
Oxford: Oxford University Press.
Thompson, S. A. (2002). ‘Object complements’ and conversation: Towards a realistic account.
Thompson, S. A., & Mulac, A. (1991a). A quantitative perspective on the grammaticalization
of epistemic parentheticals in English. In B. Heine and E. Traugott (eds.), Approaches to
grammaticalization, 313–329. Amsterdam/Philadelphia: John Benjamins.
Thompson, S. A., & Mulac, A. (1991b). The discourse conditions for the use of the complemen-
tizer that in conversational English. Journal of Pragmatics 15 (3), 237–251.
https://doi.org/10.1016/0378-2166(91)90012-M
Wang, W. (2017). Prosody and functions of discourse markers in Mandarin Chinese conversation:
The cases of ranhou, wo juede, and meiyou. Doctoral dissertation, University of California,
Los Angeles.
Part II
Locally emergent clause-combining patterns

Chapter 7
Practices of clause-combining
From complex wenn-constructions to insubordinate
(‘stand-alone’) conditionals in everyday spoken
German1
Susanne Günthner
Traditionally, conditional wenn-clauses in German are treated as subordinate

clauses, either preceding or following their matrix clauses. My data – based on
naturally occurring German talk-in-interaction from various settings – show
that participants in everyday interactions use various types of (pre-positioned)
wenn-constructions as resources to accomplish social activities. These construc-
tions not only blur the boundaries between subordinated and main clauses, but
also reveal a wide range of in-between wenn-constructions ranging from tightly
integrated to loosely integrated, from non-integrated to free-standing.
The empirically based analysis, furthermore, shows that practices of
clause-combining turn out to be closely connected to the temporal unfolding of
talk-in-interaction, and thus to issues of projection, retractive expansions as well
as the ongoing accomplishment of social action.
Keywords: emergence, co-construction, coordination, insubordination,

projection, routinization, sedimentation, stand-alone wenn-clauses,
subordination, temporal unfolding
1. Thanks to Wolfgang Imo and the anonymous reviewers for helpful comments on earlier ver-
sions. Thanks to Lisa Roebuck for checking the use of English.
186 Susanne Günthner
“Let me preface these remarks with the (hopefully unprovocative) proposition

that interactive language is the core phenomenon to be explained – all other forms
of discourse are, however interesting, derivative in every sense, ontogenetic and
phylogenetic included.” (Levinson 2006, p. 85)
1. Introduction
Over the last 10 to 15 years, interactionally-oriented approaches such as “on-line

syntax” (Auer, 2009a), “emergent grammar” (Hopper 1987, 1988; Auer & Pfänder
2011), “dialogical grammar” and “dialogic syntax” (Linell 2009, Günthner, Imo
& Bücker 2014; Du Bois 2014) have created new perspectives on the analysis of
grammatical structures as they unfold in real time and close attunement with co-
participants’ reactions (cf. Auer 2000, 2005, 2009a; Günthner & Imo 2006; Hopper
2011; Auer & Pfänder 2011; Imo 2013; Deppermann & Günthner 2015). These
studies have led to a radical change in the conceptualization of language. Instead
of treating grammar as a mental construct which speakers instantiate in their
spoken and written sentences, they are recontextualizing grammar as it is used in
the dialogical process of interaction to effect social actions.
In studying pre-positioned wenn-clauses in everyday spoken German,2 this
paper aims to contribute to the discussion on practices of clause-combining, by
showing how speakers use differing degrees of clausal complexities and clausal
integration as communicative resources to accomplish various communicative
projects. Further, I hope to stimulate the ongoing debate on “insubordination”
(Evans, 2007); in particular on stand-alone wenn-clauses.
Conditional wenn-constructions are generally treated as compound sentences,
consisting of a subordinate wenn-clause and a main clause. However, much of the
work done on wenn-clauses has been grounded in written texts or introspectively
obtained examples. In such data, the diversity and “openness” (Hopper, 2004) of
pre-positioned wenn-constructions as well as the interactional practices for using
these grammatical constructions are largely absent. Through empirical analysis, I
will show that beyond the standardized biclausal pattern, participants in everyday
interaction use a variety of wenn-constructions with different degrees of syntactic
and prosodic integration to realize various social activities. The findings will shed
light on the dialectic process of local emergence and the sedimentation of linguis-
tic patterns: On the one hand, participants in interaction orient to sedimented
clausal patterns (such as bi-partite wenn-gestalts) which make projection and
2. Cf. Auer (2000) and Günthner (2012) on fragmental, non-clausal wenn-units (like “wenn,
dann ja” or “ja wenn”).
Chapter 7. Practices of clause-combining 187
collaborative productions of syntactic gestalts possible in the first place. On the

other hand, the actualization of clause-combining patterns takes place in the ‘hic
et nunc’ of the ongoing interaction, responding to local contingencies and other
interactional factors.
The study is based on 70 everyday interactions (30 to 180 minutes in length),
collected from 1989 to 2014 throughout Germany: They include informal face-
to-face and telephone interactions among friends and family members, radio
phone-in calls, video-taped conversations of university office hours, data from
television talk-shows and from a reality-TV series. From this material, I randomly
collected a database of 80 pre-positioned and stand-alone wenn-constructions
and examined them for interactional and grammatical features. For this article, I
have chosen a set of illustrative examples for a detailed study of how participants
configure different types of wenn-constructions in everyday interaction.3
2. Wenn-constructions in everyday German interactions
In spoken German interactions, the standardized biclausal sentence pattern with

tight syntactic integration of the wenn-clause into its adjoining main clause (as
in Excerpt (1)) represents one of several ways of organizing wenn-constructions.
Even though the wenn-clauses in this biclausal construction can be placed before
or after the main clause they modify,4 studies of conditional clauses in German
show that the wenn-unit (the protasis) more commonly occurs in initial position,
projecting a subsequent main clause (the apodosis) (Auer 2000; Auer & Lindström
2015; Günthner 1999, 2012, i. pr.; Wegner 2010).5
Three friends (Janina, Birte and Markus) are talking about regionally different
German words for carrots:
3. Cf. Bergmann (1992) and Deppermann (1999) on qualitative methods and questions of
sampling in CA.
4. In rare cases, the wenn-unit can even be inserted parenthetically (interrupting the produc-
tion of the main clause); cf. Auer (2000).
5. Whereas 58% of the wenn-clauses in my data are pre-positioned, 31% are post-positioned.
11% are either stand-alone wenn-units, wenn-fragments or unclear cases (false starts, reformu-
lations, etc.). Greenberg (1963, pp. 84–85) even treats this ordering of an initial if-clause and a
following consequence clause as “universal word order”, stating: “In conditional statements, the
conditional clause precedes the conclusion as the normal order in all languages.” Cf. Ford (1993)
for similar observations in English and Laury (2012) in Finnish data.
(1) Different words for carrots (MÜNSTER)6

006 Bi: also DU sagst das jedenfalls ne?
well you say-prs.2sg this anycase ptcl
well in any case you use that word right?

007 manchmal [kaROTten. ]
sometimes carrots
sometimes [carrots. ]
008 Ja: [ich verwende] BEIdes [gern?]
I use-prs.1sg both like-adv
[ I like to use] both [actually?]
009 Bi: [ach. ]
ptcl
[oh.]
010 aHA?
ptcl
well?
011 (2.0)
012 Ja: also wenn wenn man HÄSchenwitze erzählt-
ptcl if if one rabbit-dim-jokes tell-prs.3sg
well if if you’re telling rabbit jokes-
013 spricht man doch immer von MÖHren7.
speak-prs.3sg one ptcl always of carrot-pl
you tend to use the word carrots.
014 (1.5)
015 Bi: [MÖHRchen.]
carrot-pl.dim
[little carrots.]
016 Ma: [MÖHRchen?]
carrot-pl.dim
[little carrots?]
In line 012 Janina uses a pre-positioned wenn-clause also wenn wenn man
HÄSchenwitze erzählt- ‘well if if you’re telling rabbit jokes-’ to introduce the
condition under which the proposition expressed in the subsequent main clause
spricht man doch immer von MÖHren. ‘you tend to use the word carrots’ (see l.
013) is probable. In cognitive linguistics, if-clauses are seen as setting up a “mental
space” (Fauconnier 1985; Dancygier & Sweetser 2000) and providing the back-
ground or context against which the main clause that follows is to be interpreted
6. The transcripts are based on GAT 2, a system for transcribing talk-in-interaction (Selting
et al. 2009; cf. the English translation by Couper-Kuhlen & Barth-Weingarten 2011). The tran-
scripts start with the original German utterance, followed by an interlinear gloss (based on GAT
2 and Leipzig Glossing Rules 2015) and a free translation into English.
7. “Möhren” and “Karotten” are different terms used for “carrots” in German.
or understood. The wenn-clause also wenn wenn man HÄSchenwitze erzählt- ‘well
if if you’re telling rabbit jokes-’ thus, sets up the mental space for the subsequent
proposition.
From an on-line perspective, Janina’s syntactic project in line 012 begins
with the discourse marker also followed by a wenn-clause (also wenn wenn man
HÄSchenwitze erzählt- ‘well if if you’re telling rabbit jokes’, see l. 012), which does
not constitute a self-contained unit, but projects, on various levels, a unit to come.
The information in the wenn-part provides the background (the protasis) for
the construal of the main clause (i.e., the apodosis): spricht man doch immer von
MÖHren. ‘you tend to use the word carrots.’ (see l. 013). Only with the produc-
tion of this second unit is the biclausal conditional construction complete. The
continuing intonation at the end of the wenn-clause indicates a strong syntactic
integration of the two clausal units: The wenn-clause shows subordinate word
order (i.e., final positioning of the finite verb erzählt ‘tells’) and forms the first con-
stituent (positioned in the so-called “Vorfeld” (‘front-field’) within the topology
of the sentence pattern.8 The foreshadowed main clause starts with the finite verb
spricht (‘speaks’) and thus shows ‘inversion’ (cf. verb-first positioning), indicating
a tight connection to the preceding subordinate clause.
The topological field schema displays this tight connection:
Table 1. Excerpt (1) DIFFERENT WORDS FOR CARROTS (MÜNSTER) – topological

field schema
Vorfeld linke Satzklammer Mittelfeld rechte Satz- Nach-feld
front-field left brace middle-field klammer end-field
right brace
also wenn wenn spricht man doch immer
man HÄSchenwitze speak-prs.3sg von MÖHren.
erzählt- one ptcl always of
ptcl if if one carrot-pl
rabbit-dim-jokes
tell-prs.3sg
As the data reveal, this standardized biclausal pattern is but one type of wenn-
constructions, used in everyday interactions. Speakers frequently make use of
8. Word order in German can be described by means of the so-called “sentence brackets” and
the “topological fields”: The sentence brackets, which are formed by the finite and the non-
finite verbal parts, divide the sentence into different topological fields (Zifonun et al. 1997; Auer
1996). These topological fields include the “front-field” (‘Vorfeld’) (the field before the finite
verb), the “middle-field” (‘Mittelfeld’) (the field between the finite and non-finite parts of the
verb) and the “end-field” (‘Nachfeld’) (the field after the non-finite part of the verb; this field
often remains empty).
further variants with differing degrees of integration between the pre-positioned

wenn-part and the succeeding unit.
2.1 Syntactic integration of the pre-positioned wenn-clause and the

following main clause
Even though initial wenn-clauses in my data tend to be phrased as separate

prosodic units (see Excerpt 1) with the wenn-clause ending with continuing
intonation (by means of level or rising boundary tones),9 speakers also produce
wenn-constructions in which the two clauses are tightly prosodically linked:
(2) FOOD POISONING (SCHWARZWALD)10
229 Dora: =nja und no liegts natürlich au an dir,
well and then lie-prs.3sg_it of course also on you-dat.sg

=well and then of course it also depends on you,
230 und deiner konʔ kons-
and your con cons
and your conʔ cons-
231 konstituTION;
constitution
constitution;
232 Wolf: was du verTRÄGST ja?=
what you bear-prs.2sg ptcl
whatever your body can take right?=

233 Dora: =<<all> wie gsund du grad drauf bisch;
how healthy you right_then on_top be-prs.2sg;
=<<all> how healthy you are right then;
234 und so.>
and so
and things like that.>
235 Gisa: < mhm.>
236 Udo: ja sicher naTÜRlich.
ptcl sure of_course
yeah sure of course.
237 (2.5)
238 Dora: <<all> wenn du ANGSCHT hasch kriegsch=s
sowieso eher.>
if you fear have-prs.2sg get-prs.2sg it anyway more_likely
<<all> if you are afraid you are more likely to get it anyway.>
9. Cf. also Auer/Lindström (2015, p. 7).
10. These participants use their Swabian-Alemannic dialect, which differs morphologically
from Standard German.
239 Udo: aber sag mer mol SO?

but say-prs.3sg one ptcl so
but let’s put it this way?
240 wenn sag mer mal so,
if say-prs.3sg one ptcl so
if let’s say,
241 i möcht ähm was i zum beispiel NET mache würd,
I want-prs.1sg ptcl what I for example not make-inf would-prs.1sg
I want um what I for example would never do,
Dora in line 238 uses a pre-positioned wenn-clause wenn du ANGSCHT hasch

(‘if you are afraid’; l. 238) to introduce the condition under which the proposition
expressed in the following main clause kriegsch = s sowieso eher. > (‘you are more
likely to get it anyway.>’) is likely. Both clauses – combined to express a worldly
maxim – show a strong syntactic integration (subjunction plus final positioning
of the finite verb in the wenn-unit and inversion in the main clause), which is
strengthened by the prosodic integration of the two clauses to form a single in-
tonation unit. The tight syntactic and prosodic linking of this wenn-construction
reflects the sedimented character of the presented maxim.
2.2 Syntactic (and prosodic) non-integration of the pre-positioned wenn-

clause and the following main clause
Tight syntactic integration of the wenn-clause into its adjoining main clause (see
Section 2.1) is one in a number of ways of organizing wenn-constructions. Speakers
in everyday interactions however, often make use of wenn-constructions which
deviate from this standard word order (a phenomenon less frequent in written
texts; Günthner 1999, 2012i. pr.). The pre-positioned wenn-clause can be followed
by a fully-fledged main clause which shows no inversion, and thus, does not rec-
ognize or respond to its preceding subordinate wenn-clause.11 Prosody supports
this feature of syntactic independence in so far as syntactically non-integrated
constructions are delivered with two separate intonation contours, each having its
own nucleus accent.
In our data, speakers make use of two types of syntactic non-integration:
a. non-integrated constructions which can be converted to exhibit integrative
word order
b. non-integrated constructions which cannot be converted to exhibit integra-
tive word order.
11. Various grammars mention these syntactic exceptions; cf. Zifonun et al. (1997, p. 2290) on
moduskommentierende Konditionalsätze (‘modus annotating’).
2.2.1 Non-integrated constructions which are convertible into integrative word

order
Simon (Si), Lisa (Li), Bianca (Bi) and Robert (Ro) are having dinner together when
Simon tells the others about a sushi dinner he had recently had with two other
friends (Michael and Timo). Both friends had fallen sick right after eating the
sushi. In the Excerpt (3), Lisa, Bianca, Robert and Simon are puzzling over pos-
sible reasons for their distress:
(3) EATING SUSHI (MÜNSTER)
044 Si: der hat halt AUCH-
he [det-art.nom.sg.m] have-prs.3sg ptcl ptcl
well he has also-

045 (1.0)
046 der hat halt schnell geGESsen?
he [det-art.nom.sg.m] have-prs.3sg ptcl fast ptcp-eat-ptcp
he just ate too fast?

047 und VIEL zu viel sojasoße.
and much too much soya_sauce
and used far too much soya sauce.
048 (1.5)
049 Bi: HAhaha-
050 Si: der HAT-
he [det-art.nom.sg.m] have-prs.3sg
he has-
051 Bi: Overkill an sojasoße.
overkill at soya_sauce
overkill with soya sauce.
052 ((B und L lachen))
((B and L are laughing))
053 Si: ich verSTEH das aber auch immer nich-
I understand-prs.1sg this but also always not
I still don’t ever really get it-
054 michael und timo BEIde.
michael and timo both
michael and timo both of them.
055 wenn man mit denen SUshi essen geht-
if one with those-pro.3.pl sushi eat-inf go-prs.3sg
if you go and eat sushi with them-
056 die BAden immer-
they-det-art.nom.3.pl bath-prs.3pl always
they always drown-
057 das SUshi komplett in dieser sojasoße-
the sushi completely in this soya sauce
their sushi completely in this soya sauce-
058 bis sich der ganze reis vollgesogen hat-

till itself-refl the complete rice full_suck-ptcp have-prs.3sg
till the rice is completely drenched-
059 Bi: bis der ganz SCHWARZ is.
till it-det-art.nom.sg.n totally black be-prs.3sg
till it is all black.
060 Si: ja.
yeah
yeah.
061 Li: ja aber dann schmeckt man doch gar nichts
mehr von dem REST?
yeah but then taste-prs.3sg one
ptcl at_all nothing anymore from the
rest
yeah but then you can’t taste the rest at all anymore?
Simon’s utterance in line 055 starts with an initial wenn-unit (wenn man mit denen
SUshi essen geht- ‘if/when you go and eat sushi with them’) which does not consti-
tute a syntactically or semantically complete unit nor a complete communicative
action, but gives rise to the expectation of another component – the consequence.
The two units of this syntactic gestalt are uttered as separate intonation units, each
with its own independent intonation contours and nucleus accent. The wenn-part,
introducing a story, is clearly marked as a subordinate clause (by the initial subjunc-
tion wenn and final placement of the finite verb). It ends on a level tone, indicating
‘continuation’. The action reconstructed in the main clause (die BAden immer- das
SUshi komplett in dieser sojasoße- bis sich der ganze reis vollgesogen hat- ‘they
always drown- their sushi completely in this soya sauce- till the rice is completely
drenched-‘, see l. 056–058) is presented as valid, should the condition stated in the
wenn-clause (the protasis) apply. However, despite the dependent clause features
of the wenn-unit, its adjoined main clause fails to display inversion (verb-first-
positioning). Instead, the consequence die BAden immer- das SUshi komplett in
dieser sojasoße- bis sich der ganze reis vollgesogen hat- ‘they always drown- their
sushi completely in this soya sauce- till the rice is completely drenched-’ (see l.
056–058) is attached asyndetically without showing the grammatical coding ex-
pected as a result of the preceding subordinate wenn-clause. Instead the complex
consequence unit – comprising three TCUs with tension-holding level final pitch
movements – gradually unfolds the foreshadowed punchline of the story.
The grammatical cohesion between the two clauses is weakened, and the
subordinate wenn-clause no longer occupies the position of the front field, but is
shifted to the peripheral position of the “Vor-Vorfeld” (‘pre-front field’; Auer 1996):
Table 2. Excerpt (3) EATING SUSHI – topological field schema

Vor-Vorfeld Vorfeld linke Satzklammer Mittelfeld rechte Satz-klammer Nachfeld
pre-front-field front-field left brace middle field right brace end-field
wenn man mit denen die BAden immer- das sUshi bis sich der ganze
sUshi essen geht- komplett in dieser Reis vollgesogen hat-
sojasoße-
if one with them sushi they -det-art. bath-prs.3pl always the sushi till itself-refl the
eat-inf go-prs.3sg- nom.3pl completely in this soya complete rice full_suck-
sauce ptcp have-prs.3sg
Even though the main clause depicts a syntactically asyndetic unit, this construc-
tion can be converted into a syntactic pattern showing integrative word order:
wenn man mit denen SUshi essen geht- BAden die immer- das SUshi komplett in
dieser sojasoße- bis sich der ganze reis vollgesogen hat. From the perspective of
on-line syntax, one might argue that until the end of the wenn-unit, speakers have
the opportunity to present the wenn-construction as syntactically integrated (with
the main clause showing inversion), or as two loosely connected units with the
main clause conjoined asyndetically.
In contrast to integrated wenn-clauses, syntactically non-integrated ones tend
to be used in contexts in which the initial wenn-part works to organize the on-line
management of interactional contingencies by framing the social activity which
follows. Here, Simon implements the wenn-part to build up the scene for the fol-
lowing ‘climax’ (lines 056–058). Due to the non-integrated syntax, the projected
consequence shows greater freedom in its syntactic design. This flexibility may
be used (as in this excerpt) for staging the syntactically as well as prosodically
highlighted punchline illustrating his friends’ bizarre way of eating sushi.
2.2.2 Non-integrated constructions which are not convertible into integrative

word order
In contrast with the wenn-construction in Excerpt (3), participants also use syn-
tactically non-integrated wenn-constructions, which cannot be reformulated as
integrated ones without changing their meaning.
In Excerpt (4), Ute comments that she is helping herself to an apple lying on
Michi’s kitchen table (see l. 288). Two seconds after Michi welcomed Ute’s course
of action (see l. 289), she introduces a subordinate wenn-clause wenn er dir net
SCHMECKT, ‘if you don’t like it,’ (see l. 291), followed by an asyndetically orga-
nized main clause mir hän no ANdere. ‘we have other kinds [too.]’ (see l. 292):
(4) APPLES (BODENSEE)
288 Ute: <<☺> ich NEHM mir einen?>
I take-prs 1. sg me-dat.sg one
<<☺>I’ll have one?>
289 Michi: KLAR (gern.)
sure (willingly-adv)
sure (you are welcome to.)
290 (2.0)
291 wenn er dir net SCHMECKT,
if he-GENUS-M to you not taste-prs 3.sg
if you don’t like it,
292 mir12 hän no13 ANde[re.]

we have-prs 2.pl still others
we have other kinds [too.]
293 Ute: [öhm] ne=ne.
ptcl no no
[um] no=no.
294 <<:-)> i- ich mag so sAUre ganz GERN.>
I- I like-prs.1.sg so sour.pl quite a_lot
<<:-)> I- I do actually like the sour ones quite a lot.>
The initial wenn-unit in line 291 has all the indications of a subordinate clause.
However, in spite of the dependent clause features of the wenn-unit, its adjoined
main clause fails to display inversion (i.e., verb-first-positioning). Prosodically,
the wenn-unit, which suggests the possibility that Ute might not like this apple,
is realized as a self-contained contour, even though its rising boundary tone
implies ‘continuation’. What follows in the second unit (mir hän no ANdere. ‘we
have other kinds [too.]’, see l. 291) is not a direct consequence of the statement
in the protasis (wenn er dir net SCHMECKT, ‘if you don’t like it,’, see l. 291), but
an offer, which holds in case Ute does not like her apple. The information mir
hän no ANdere. ‘we have other kinds [too.]’ is valid, whether or not Ute likes her
apple. Hence, this construction carries a “non-conditional meaning” with the pre-
positioned wenn-clause forming the background against which the subsequent
offer may become relevant.14
As in Excerpt (3), the grammatical cohesion between the two clauses is re-
duced and the subordinate wenn-clause occupies the peripheral position of the
“Vor-Vorfeld” (‘pre-front –field’) and thus the topological field used in spoken
German to frame subsequent utterances (Auer, 1996, p. 310)15:
The reduced syntactic tying between the wenn-part and the following (seem-
ingly independent) syntagma reflects the “metapragmatic function” (Silverstein
1993; Maschler 2009) of the initial wenn-unit: The condition ‘if you don’t like it,’
(see l. 291) “prepares the scene” (Schegloff, 1984), under which the foreshadowed
activity can be relevant for the recipient. Thus, the syntactic architecture of the
clause-combining iconically reflects the different interactional levels the two parts
of this compound construction are operating on.
12. “mir” is the Swabian-Alemannic dialect term for wir ‘we’.
13. “no” is the Swabian-Alemannic dialect term for noch ‘still’.
14. Cf. Sweetser (1990) and Günthner (1999, 2009, 2012) for so-called “relevance conditional”.
15. In spoken German, the general function of the pre-front field is to frame subsequent utter-
ances (Auer, 1996, p. 310); framing functions of wenn-clauses include pragmatic specifications,
modalizations, discourse-organizational aspects etc.
Table 3. Excerpt (4) APPLES (BODENSEE) – topological field schema

Vor-Vorfeld Vorfeld linke Mittelfeld rechte Satz- Nachfeld
Satzklammer klammer
pre-front-field front-field left brace middle-field right brace end-field
wenn er dir net mir hän no ANdere.
SCHMECKT,
if he-GENUS-M to you not we have-prs 2.pl still others

taste-prs 3.sg
In studying wenn-constructions as they emerge in the process of interaction,

we can detect striking parallels between the syntactically non-integrated wenn-
clauses and “projecting constructions” (Günthner 2006, 2008a, b, 2011a, b; Hopper
& Thompson 2008; Günthner & Hopper 2010; Pekarek Doehler 2011) such as
N-be-that-constructions, pseudoclefts, etc., organizing the on-line management
of interactional contingencies and preparing for the particular social activities that
may follow. Positioned in the pre-front-field, the wenn-unit functions as a framing
device and “prepares the scene” (Schegloff, 1984) for the subsequent activity. The
syntactic and prosodic disintegration – and thus, the loose connection between
these clauses – iconically mirrors the different levels of activities (metapragmatic vs
pragmatic).16 However, in spite of the loose syntactic (and prosodic) connection be-
tween the two parts, the wenn-clause – by means of its projecting force – functions
as an important interactive resource speakers use for claiming their right to keep the
turn till the end of the construction. This turn-keeping function is particularly rel-
evant in cases in which speakers produce longer stretches of discourse (Section 2.4).
2.3 Co-constructed wenn-constructions
One of the main characteristics of everyday interaction is its inherent dialogic-

ity (Luckmann 1990; Linell 2009; Günthner 2015a, b) and thus, ego’s orientation
towards alter-ego (Schütz 1932/1973: 66). As Luckmann (1990, p. 52), following
Schütz, argues: “People participating in dialogue are in the presence of one an-
other, they are in the unique position of sharing time and space with one another.
Sharing a sector of the world is a prerequisite for mutual attention and a condition
for what Alfred Schütz called “the synchronization of two streams of conscious-
ness” Schütz 1932/1973: 123.
Due to this synchronization of the participants’ minds in face-to-face-
interaction, speaker and hearer become closely attuned and aligned: They attend
to the ongoing process of interaction, adjust their contributions to the sequential
16. Cf. Maschler (2009).

organization in time and collaborate in the emergence of communicative projects

in social interaction. The inherent dialogicity and synchronization of participants’
streams of consciousness have (as Excerpt (5) will show) significant effects on the
production of grammatical structuring, with participants working together on the
moment-by-moment unfolding of grammatical gestalts. Their shared orientation
towards the dynamics of the interactive process and, thus, their close attunement
to each other’s contributions can be observed most vividly in the collaborative
instantiation of the complex wenn-construction in Excerpt (5). Rolf and Lea are
visiting their mother and Rolf proposes to cook dinner the next day. Talking about
what to cook, he produces the following wenn-clause: °hh wenn_s gnug mEhl, eier
on:d eh KÄS gäb, ‘°hh if there was enough flour, eggs and cheese,’ (see l. 221). Right
after the completion of this initial wenn-part, Lea takes over and utilizes Rolf’s
wenn-clause as an anchor for her syntactic continuation (the apodosis) (see l. 222):
(5) DINNER (SCHWARZWALD 1998)
219 Rolf: [joah: (aber)] was soll i KOCHe?
yeah (but) what shall-prs.1sg I cook-inf
[yeah: (but)] what shall I cook?
220 (1.5)
221 °hh wenn_s gnug mEhl eier on:d eh KÄS gäb,
°hh if_it enough flour eggs and eh cheese have-subj.3sg
°hh if there was enough flour, eggs and eh cheese,
222 Lea: <<lachend> d_dann dÄtsch KÄSspätzle
[mache. >]
th_then do-subj.2sg KÄSspätzle make-inf
<<laughingly> th_then you’d prepare [KÄSspätzle17.>]
223 Rolf: [(ja klar)]
(yeah sure)
[(yeah sure)]
224 Ulla: haJO.
of_course
of course.
Rolf and Lea co-construct a syntactic pattern, consisting of two syntactically and
semantically connected clauses. Introduced with the connective dann (‘then’),18
Lea nests into Rolf’s initiated syntactic gestalt by providing the projected main
clause and, thus, closes the complex syntactic gestalt. Starting her turn with dann
(‘then’), Lea indicates that the following unit will present a continuation, providing
the expected apodosis to first speaker’s protasis.
17. “Kässpätzle” is a local dish, consisting of Swabian noodles with cheese and onions.
18. Dann (‘then’) in wenn-constructions is generally treated as a “resumptive particle”, as it

“resumes a conditional antecedent” (König & Van der Auwera 1988: 121).
Co-produced wenn-constructions, stretching over two different speaker’s

turns, not only shed light on the emergent production of a syntactic construction,
but also provide “live” evidence for participants’ attending to commonly shared
syntactic gestalts.
Even though second speakers may align with first speaker’s initiated syntactic
project, they may complete the projected grammatical gestalt in a discordant way
and introduce a disaffiliating stance. I.e., a second speaker may join the ongo-
ing syntactic project, but at the same time introduce her/his disagreeing stance
towards first speaker’s position.
Tim and his cousins Ina and Anni are talking about Tim’s problems with his
girlfriends. In line 035, Tim initiates the first part of a complex wenn-construction;
however, immediately after his introduction of the foreshadowed consequence
by means of the connective no: (Swabian for dann; i.e., ‘then’), Anni takes
over (see l. 039):
(6) NEW GIRLFRIEND (SWABIA 1996)19
031Tim: WEISCH- (-)h°
know_you-prs.2sg h°
you know- (-) h°
032 i i brAUch mein FREIraum, (-)
I I need-prs.1sg my free_space
I I need my space, (-)
033 [ ECHT. ]
really
[for sure.]
034 Anni?: [( ) ]
035 Tim: on wenn ich den net krieg,
and if I it-genus-m.sg not get-prs.1sg
and if I don’t get it,
036 h° on muss mi emmer (.) emmer rechtFERtige,
h° and must-prs.1sg myself always (.) always justify-prs.1sg
h° and always (.) always have to justify myself,
037 wenn i mol mit ANdere was mAche will,
if I ptcl with others something make-inf want-prs.1sg
whenever I want to hang around with other people,
038 [no: ]20
then
[then:]
19. Cf. also Günthner (2015b).
20. no: is the Swabian dialect term for dann (‘then’).

039 Anni: [dann] <<f> verLÄSCH se,

then leave-prs.2sg her
[then] <<f> you’re going to leave her,
040 und fÜrsch de uff,
and act-prs.2sg yourself ptcl
and act,
041 wie [<<☺> en alter MATSCHo.>>]
like an old macho
like [<<☺> an old macho.>>]
042 Ina: [ ((lacht)) ]
[ ((laughs)) ]
043 Ina&Anni: [((lachen)) ]
[((laughter))]
044 Tim: [No ]
no
[no ]
045 (von wege) mer isch net glei en MATscho,
(as if) one is-prs.3sg not right_away a macho
(as if) one turns into a macho,
046 b_ bloss (.) wenn mer sein FREIraum [braucht.]
o_ only if one his free_space need-prs.3sg
o_ only (.) if one needs one’s own [space. ]
Excerpt (6) shows how co-participants monitor speaker’s ongoing syntactic forma-
tions and how they (here: Anni) – at just the right moment – step in and ‘take the
projected continuation out of speaker’s mouth’ by completing the initiated syntac-
tic construction. Starting her turn with dann (l. 039),21 the second speaker marks
her utterance as a continuation of the preceding unit and thus, as the consequence
of Tim’s statement. Though Anni’s turn [dann] < <f > verLÄSCH se, und fÜrsch
de UFF, wie [<<☺> en alter MATSCHo.>>] ‘[then] <<f> you’re going to leave her,
and act, like [<<☺> an old macho.>>]’ (see l. 39–41) provides the syntactically
projected continuation, it carries a disaffiliative stance. Despite Anni’s formally
aligned continuation, she teasingly – contextualized by the smile voice – attributes
an intention to first speaker which clearly stands in contrast to his intention. Ina’s
laughter displays her appreciation of Anni’s teasing critique and of the sudden
turnabout in the interactive process.
Lines (5) and (6) not only demonstrate how participants in the emergence of
talk-in-interaction co-construct a common syntactic gestalt, but they also indicate
how intersubjectively shared grammatical knowledge (Günthner 2014, 2015a, b, i.
21. Collaboratively constructed wenn-clauses may show syntactic integration as well as non-
integration. Frequently, the consequence introduced by the second speaker starts with dann,
marking a close connecting to first speaker’s antecedent. Cf. also Günthner (2015b), i. pr..
pr.; Hilpert 2015) is reactivated in the ongoing process of interaction.22 Jointly pro-
duced wenn-constructions form insightful in-situ indicators for the interrelation
between shared and sedimented constructional knowledge and the emergence
of grammatical structuration (Hopper, 1987): They are not achievements of an
individual mind, but form coordinated practices and accomplishments negoti-
ated between co-participants when constructing the patterned orderliness of
everyday interaction.
2.4 Beyond biclausal sentence patterns: Wenn-constructions comprised of

longer discourse units
As Excerpts (7) and (8) will show, participants implement wenn-constructions to

claim the turn for longer stretches of discourse. This turn-holding function is used
for various social activities, e.g. providing expanded advises and teaching activi-
ties, argumentative sequences, as well as other “big packages”.
2.4.1 Incrementally expanded wenn-units, spreading over several TCUs

Speakers frequently use multi-unit wenn-parts incorporating a string of different
clauses which stretch over several turn construction units (TCUs) and so, consti-
tute an emerging sequence of talk.23
Jan and his niece Mia are talking about the early onset of winter, when Jan
urgently advises Mia against using her bike to ride to school. His wenn-unit (l.
76–90) stretches over several TCUs occupying an extended sequential space which
includes various dass-clauses, parentheses, lists, main clauses, etc.:
(7) ONSET OF WINTER (MÜNSTER 2009)24
74 Jan: als jemand der STÄNDig damit zu tUn hat,
as someone who constantly with_that to do-inf have-prs.3sg
as someone who constantly has to deal with such issues,
75 SACH ich dir jetz mal wAs; (0.3)
say-prs.1sg I you-dat now ptcl something
let me tell you something now; (0.3)
76 h° wenn DU meinst-
h° if you think-prs.1sg
h° if you think-
22. Cf. also Lerner (1991, 2002) on collaborative turn-sequences and syntactically co-con-
structed sentences.
23. See also the multi-unit wenn-part in Excerpt (6); l. 35–37). Cf. also Wegner (2010).
24. Cf. also Günthner (2015b).

77 du hättest d dein rAd unter kontROLLe;

you have-subj.2sg y your bike under control
you are in control of y your bike;
78 dass du BREMsen kannst,
that you brake-inf can-prs.2sg
that you can apply the brakes,
79 dass du [auch] vorʔ LANGsam fÄhrst,
that you [also] care- slowly ride-prs.2sg
that you [also ] ride careʔ slowly,
80 Mia: [ja; ]
yeah
[yeah; ]
81 Jan: dass du AUf[passt ] wo_s GLATT is;
that you ptcl-[take-prs.2sg] where_it icy be-prs.3sg
that you are [careful ] where it’s icy;
82 Mia: [< ja;>]
yeah
[<yeah;>]
83 Jan: un ähn dass du (.) MEINST,
an ptcl that you think-prs.2sg
an um that you (.) think,
84 du hast alles im GRIFF;
you have-prs.2sg everything in hold
that you have everything under control;
85 sowas pasSIERT dir nich;
such-a-thing happen-prs.3sg you-dat not
things like that can’t happen to you;
86 das pAssiert nur solchen DÖ:deln da,
this happen-prs.3sg only such-dat fools-dat there
they only happen to fools,
87 dir aba [ nIch; ]
you-dat but not
but not to [you; ]
88 Mia: [(nein;)]
no
[(no;) ]
89 Jan: da tÄUscht du dich g ganz geWALTich;
there deceive-prs.2sg you yourself v very tremendously
then you are s seriously mistaken;
90 du kAnnst bei der glätte nich BREMsen;
you can-prs.2sg at this iciness not brake-inf
you just cannot brake when it’s icy;
91 dein rad kommt ins SCHLEUdern; (.)
your bike come-prs.3sg into skidding
your bike is going to skid; (.)
92 h° äh und dann hAste den SCHEISS.
h° ptcl and then have-prs.2sg the shit
h° um and then you’re in deep trouble.
The temporal unfolding of Jan’s wenn-construction illustrates that this wenn-clause

(l. 76–90) does not represent a syntactically or semantically complete unit, nor a
complete communicative action; but it does call for a further component – the
consequence (i.e., apodosis). However, this communicative project is not com-
pleted right away: The subsequent complement clauses (l. 78ff.) dass du BREMsen
kannst, dass du [auch] vor- LANGsam fÄhrst, dass du AUf[passt] wo_s GLATT
is; […] ‘that you can apply the brakes, that you [also] ride care- slowly, that you
are [atten]tive to slippery ground; […]’ do not satisfy the wenn-clause projection
impetus, but they incrementally expand the wenn-part and extend the projection
span, thus keeping the projecting force in play. In line 83, Jan stages Mia’s possible
thoughts and opinions on the matter at hand over an extended turn by using a
conjoined construction: un ähn dass du (.) MEINST, du hast alles im GRIFF; sowas
pasSIERT dir nich; […] ‘an um that you assume, that you have everything under
control; such things can’t happen to you; […]’. Once more, the speaker does not
fulfill the projection, so the projecting force of the wenn-clause in line 76 remains
in effect. Jan’s subsequent utterance: das pAssiert nur solchen DÖ:deln da, dir aba
[nIch;] ‘they only happen to fools’, which continues the inner monologue and
confronts Mia with her possible thoughts, further delays the completion of the
wenn-construction. Finally, the utterance in line 89, introduced with the adverb
da ‘there/then’, links back to the syntactically projecting wenn-component in line
76 (da tÄUscht du dich g ganz geWALTich; du kAnnst bei der glätte nich BREMsen;
‘then you are seriously mistaken; you just cannot brake when it’s icy;’, see l. 89–90).
This segment illustrates how the wenn-construction emerges as a solution to
the real time unfolding interactional contingencies the speaker (Jan) meets while
trying to teach his niece about the dangers of riding her bike in winter. The wenn-
clause h° wenn DU meinst- […] du hättest d dein rAd unter kontROLLe; ‘h° if you
think- you are in control over y your bike;’ (see l. 76–77) anticipates a continuation
in form of a consequence and thus guides the recipient’s attention to what fol-
lows (the speaker’s warning). At the same time, the projecting force of this unit
ensures the speaker the opportunity to finish his “big package” (i.e., his teaching
sequence) and keep the floor pending until the formulation of the foreshadowed
consequence.25 Although initial wenn-clauses may project the consequence, it
need not be presented immediately; a projection may be deferred and still retain
its validity across longer sequences of talk.26
25. Cf. Auer (2000, p. 184).
26. As Auer (2009b, p. 184) points out, empirical studies have yet to be conducted on how long
a projection can be ‘in play’ before it is forgotten by the speaker or the recipient: “It seems likely
that there is a certain time span which must not be exceeded though.”
Excerpt (7), furthermore, shows how speakers use wenn-units as a dynamic

form produced online closely attuned with their participants. Thus, the concept
of wenn-constructions as two juxtaposed clauses is not compatible with interac-
tive reality: It ignores the fact that clause-combining in everyday interactions is
an emerging phenomenon expanding in real time and closely negotiated with
speaker’s co-participants.
2.4.2 Wenn-clauses followed by a longer sequence of talk:

Not only may the wenn-parts be comprised of various clauses and intonation units,
taking up extensive time and sequential space, but the second unit (the apodosis
or rather consequence) may also embrace a longer stretch of talk whose end is
sometimes difficult to determine (Günthner 1999, 2012; Wegner 2010).27
Excerpt (8) is taken from a talk show with Heidi Klum, the producer of the
TV-show “Germany’s next top model”. The moderator (Mod) talks to Klum about
the problem of many young women starving themselves in an attempt to meet
current ideals of the so-called SCHLANKWAHN (‘slimness frenzy’), and criticizes
Klum for treating underweight girls as if they were “too fat”:
(8) THE OBSESSION WITH BEING THIN (HEIDI KLUM: 2006)28
431 Mod: ich verSTEH das alles,
I understand-prs.1sg this all
I understand all of this,
432 so wie du das beSCHREIBST,
so how you it describe-prs.2sg
the way you are describing it,
433 aba (--) äh:-
but ptcl
but (--) um:-

434 wenn man da mal jetzt REINgucken,
if one there ptcl now inside_look-inf
if we look at it more closely,
435 es gibt ja in diesem mOdelBUSiness,
it give-prs.3sg ptcl in this modelling business
in this modelling business there are,
436 VIEle mädchen die einfach WAH::Nsinnig dünn sind,
many girls who just awfully thin be-prs.3pl
many girls who are just awfully thin,
27. Wegner (2010, p. 13) found that 20% of the wenn-parts in his data showed a complex second
component. In 15% of the constructions, the wenn-part was comprised of various TCUs and in
7% both parts represented complex units.
28. Cf. Günthner (2012).

437 [es] gibt hier ein BILD,

it give-prs.3sg here a photo
[here] is a photo,
438 Klum: [mhm]
439 Mod: SCHLANKwahn; (-)
slimness_craze
slimness craze; (-)
440 WILLST du das? (.)
want-prs.2sg you this
is that what you want? (.)
441 heidi klum, (-)
heidi klum
heidi klum, (-)
442 ja es is kEIn [model aus] deina SENdung,
yeah it be-prs.3sg no model from your program
true this is not a [model from] your program,
The adversative conjunction aba ‘but’ in line 433 foreshadows an opposing argu-
ment. The moderator goes on to use a subordinate wenn-clause wenn man da mal
jetzt REINgucken, ‘if we look at it more closely,’(see l. 434), presented as a separate
intonation unit. Its rising boundary tone projects a ‘continuation’ and implies the
presentation of a consequence. However, as Auer (2009b) outlines, projection
never equals determination. This also holds for wenn-constructions: The initial
wenn-unit gives the speaker a certain amount of leeway in shaping the following
part. In this segment, the wenn-unit is neither proceeded by a direct consequence
nor is the following part limited to a single clause. Instead, the moderator’s argu-
mentation is comprised of a string of clauses, which extend over several TCUs (l.
435–440) and include declarative as well as interrogative clauses. The syntactic
disintegration, again, reflects the loose semantic connection between the units.
The wenn-clause no longer presents the condition under which the following
statement is valid, but creates a foundation for the subsequent argument, which
ends with a rhetorical question (l. 439–440). In light of the complexity of this
argumentative sequence, it is unlikely that the construction was planned in its
entirety when the initial wenn-clause was produced.29 Instead, we are dealing with
the process of an argumentative sequence unfolding over time.
Excerpts (7) and (8) illustrate that both units, the wenn-part and the follow-
ing consequence, may be realized as segments of varying lengths and complexity
extending over several TCUs with no discernible boundary and involving various
activities. Jespersen’s (1924, p. 26) statement that “…a sentence does not spring
29. Cf. Laury, Lindholm & Lindström (2013, p. 236) for similar cases in Swedish and Finnish
conditional constructions.
into a speaker’s mind all at once, but it is framed gradually as he goes on speaking”,
also holds for complex clause-combining: Syntactic constructions can be recon-
figured in response to locally managed interactional needs; i.e., they are subject
to local contingencies and time-bound social activities (Auer 2007; Günthner
& Hopper 2010; Günthner 2006, 2008b, 2011a, 2015c; Pekarek Doehler 2011;
Deppermann & Günthner 2015). Hence, clause-combining is not to be treated as
a static phenomenon, but a dialogically oriented multi-semiotic process, emerging
in the ‘hic et nunc’ of social interaction.
2.5 Stand-alone wenn-constructions
So far, I have maintained that initial wenn-clauses represent prosodically, syntac-

tically, and semantically incomplete units: They open a syntactic gestalt, which
creates the expectation of a second unit whose grammatical make-up may vary in
several ways; i.e., the second units may range from an integrated or autonomous
main clause up to a considerably longer stretch of discourse. As Auer (2000, p. 184)
points out “producing a wenn-clause gives the speaker the right and obligation to
go on talking. It functions as a turn-holding device until the formulation of the
consequence is completed.”
However, in everyday interactions, speakers not only produce wenn-units
which are (tighly or loosely) connected to their following syntactical units (cf.
Section 2.2 to Section 2.4), they also employ wenn-units which are neither fol-
lowed nor preceded by a superordinated clause. Still, these wenn-units (without
an accompanying matrix clause) constitute self-contained communicative actions
and are non-projecting (Laury et al. 2013).
“Independent verb-final clauses” or rather “isolated subordinate clauses” have
been discussed by German grammarians for quite some time (Brugmann 1918;
Oppenrieder 1989). In general, they have been associated with particular sentence
types or activities, such as “optative sentences”, expressing wishes and desires
(wenn ich doch bloß Millionär wäre!; ‘if only I were a millionaire!’) and requests
(wenn ich um eine Antwort bitten dürfte; ‘if I could ask for an answer, please’)
designed to ask a recipient to perform some action.
In the following, I will show that speakers use stand-alone wenn-constructions
for implementing various social activities. Some of these usages have become rou-
tinized practices for implementing wishes, requests or threats. However, in natu-
rally occurring interactional environments we can detect less routinized forms,
too (Günthner 2012, 2017). These findings call into question the binary division
between subordinate and main clauses (Matthiessen & Thompson 1988; Günthner
1999; Laury 2012; Laury et al. 2013) and hints at intermediate stages arising out
of the seeming contradictions between syntactically marked subordination and

pragmatically independent communicative actions.
2.5.1 Wishes and requests

Excerpt (9), taken from a radio-phone-in program, illustrates a wenn-clause,
which is not integrated in any superordinate main clause but constitutes a turn in
and of itself. The moderator (Mod) asks his caller, in the form of a wenn-clause,30
to name some celebrities whom he considers embarrassing:
(9) EMBARRASSING CELEBRITIES (RADIO-PHONE-IN #48)31
168 Mod: ehm (.) VIERte frAge,=
ptcl fourth question
um (.) fourth question,=
169 =wElchen dEUtschen prominenten findest !DU!
am pEInlichsten?
which german celebrity find-prs.2sg you the most embarrassing
=which german celebrity do you think is the most
embarrassing?
170 (---)
171 Jan: och=das würd(e) ich nie sagen jetzt.(.)
ptcl this would-1sg I never say-inf now
oh= I would never say that now. (.)
172 also d_d da kOmmen EINige;
ptcl t=t there come-prs.3pl some
well t_t there are several who might qualify;
173 da hab ich(-)da_hab ich ne GUte top tEn liste.
there have-prs.1sg I there_have-prs.1sg I a good top ten list
I have (-) actually I have a good top ten list.
174 Mod: !HACH!
ptcl
!wow!
175 Jan: ((lacht))
((laughs))
176 Mod: <<☺> wenn ich doch jetzt EIn_n NAmen
erfahrn könnte.>
if I ptcl now one name find_out-inf can-subj.1sg
<<☺>if I could just get one name now.>
30. This type of stand-alone wenn-clause comes close to what is traditionally treated as an
“optative clause”; i.e., representing a grammatical mood that indicates speaker’s wish or hope
(cf. Oppenrieder, 1989).
31. Cf. also Günthner (2017).

177 Jan: öh: NEIN.

ptcl no
uh: no.
178 das das is is doch nich,
this this is is ptcl not
this this is is not,
179 !NEIN!.
no
!no!.
180 ich will doch nich dAUernd über BOHlen32 sprechen.
I want-prs.1sg ptcl not continously about Bohlen talk-inf
I really don’t want to talk about Bohlen all the time.
The moderator’s turn in line 176 begins with an initial wenn-clause <<:-) > wenn
ich doch jetzt EIn_n NAmen erfahrn könnte.> ‘<<:-)> if I could just get one name
now.>’ (see l. 176), exhibiting typical subordinate characteristics such as an initial
subjunction and verb-late placement. Still, this seemingly subordinate syntactic
unit forms a self-contained turn and represents a fully-fledged activity: It expresses
a wish aiming at encouraging the co-participant to do something that will benefit
the initiating speaker (Couper-Kuhlen, Fox & Thompson 2014, p. 122). The modal
particle doch adds pressure to the moderator’s wish and supports the interactive
function as a request, making a response from Jan to accede to the request a relevant
next. Prosody fortifies the ‘independence’ of the wenn-clause: it is produced as a
prosodic unit of its own with utterance final prosody. The co-participant’s reaction
supports this interpretation: Jan treats this wenn-unit as an accountable action in
its own right. He responds to the moderator’s indirect request (l. 177–178) with a
moaning öh: NEIN. ‘uh: no.’, showing no indication that he is anticipating a main
clause. Subsequent to his playful fuss, Jan, in line 180, complies with the modera-
tor’s request and provides the name of a German celebrity.
In Excerpt (10) from a university office-hours session, the professor (Prof)
uses a stand-alone wenn-clause to formulate an explicit request for action:
(10) OFFICE HOUR: INTERNSHIP REPORT (VIDEO – MÜNSTER)
046 Prof: da fehlt nur noch das SIEgel.
here lack-prs.3sg only still the stamp
only the stamp is missing here.
047 Stu: achSO.
pctl
I see.
32. Bohlen is a German musician.

048 Prof: das muss ich eben-

this must-prs.1sg I just
hold on I just have to-
((the professor moves his chair back to get at his stamp))
049 Stu: hüps?
pctl
oopps?
((in doing so, the professor lets go of the report on his desk,
the page he had opened closes))
050 Prof: OH.
PCTL
oh.
051 RUTSCHT uns hier alles-
slip-prs.3sg us-refl here everything
everything is falling apart here-
052 wenn sie_s FESThalten [würden-]
if you_it hold-inf would-3pl
if you could hold [it-]
053 Stu: [joa? ] ((lacht kurz))
[yeah?] ((laughs for a moment))
054 das GEHT-
this go-prs.3sg
no problem-
After realizing that the pages of his manuscript are fluttering around, the speaker
asks his student to hold his papers. His utterance in line 052 starts with the sub-
junction wenn and carries late placement of the finite verb (würden ‘would/could’).
It thus exhibits characteristics of a canonical dependent clause. The professor’s
request, realized in the subjunctive mood (FESThalten [würden-] ‘could hold’), is
designed to get his student to perform some action (Couper-Kuhlen et al. 2014,
p. 122). In response to this request, the student agrees [joa?] das GEHT- ‘[yeah?]
no problem’ (see l. 053) and holds on to the paper. The immediacy of her reaction
suggests that she is not waiting for a continuation, but interprets the wenn-clause as
self-contained. Again, the action realized by means of the stand-alone wenn-unit
(the request) makes co-participant’s compliance expectable; and the requested
action follows straight away.
Stand-alone wenn-clauses which exhibit typical features of grammatical
subordination, but constitute prosodically and pragmatically autonomous activi-
ties, can be treated as cases of “insubordination” (Evans, 2007), and thus, “as the
conventionalized main clause use of what, on prima facie grounds, appear to be
formally insubordinate clauses” (Evans, 2007, p. 367).33
33. Cf. also Günthner (2014, 2017) on insubordinate dass- and wenn-clauses.
Single if-clauses expressing wishes, requests, proposals or suggestions turn out

to be routinized constructions in various languages. As “politeness conditionals”34
(Metschkowa-Atanassowa, 1983; Laury 2012) they are treated as forms of “indi-
rect requests”, giving the addressee room for negotiation (Laury 2012, p. 222) and
thus, soften the directive.35 However, even though the conditional architecture of
“requests for action” (Couper-Kuhlen et al. 2014, p. 122) displays the requester’s
awareness that there may be contingencies which could make it difficult to grant
the request, these constructional formats are so conventionalized that the argument
that “insubordinated requests are favoured in requests for reasons of politeness
by virtue of playing down the explicit interpersonal control made evident in im-
peratives and other indirect commands” seems to be an oversimplification (Evans
2007, p. 393). The conventionalization of this construction is already advanced to
such a degree that co-participants do not have to supplement any ellipsed mate-
rial in order to interpret it correctly. As Excerpt (10) illustrates, co-participant’s
immediate verbal and non-verbal responses indicate that she (Stu) interprets this
stand-alone wenn-construction as a fully-fledged request without hesitation.
2.5.2 Warnings/threats:
In addition to wenn-clauses expressing wishes and requests, speakers in my data
apply these seemingly subordinate clauses as resources with which to express
prohibitives; i.e., warnings and threats.
In Excerpt (11), taken from a reality-TV-show, Sabrina reconstructs an episode
in which she, as a child, nearly cut off her finger and tried to hide it from her strict
mother. The next day, however, as Sabrina’s injury worsened and the bandage was
soaked through, her mother noticed the wound and became angry with Sabrina:
34. Studies of spoken Finnish (Laury, 2012), Swedish (Laury et al. 2013), Italian (Vallauri 2004),
English (Stirling, 1999) and Dutch (Boogaart & Verheij 2013) point to the fact that stand-alone
conditionals represent conventionalized means associated with directive actions in all these
languages. Brown and Levinson (1978/1987, p. 153–154) argue that in Tzeltal, the possibility
marker “me” (‘if ’) is used to soften commands. By “including a notion of possibility or ‘if ’ in the
command”, it comes close to a “polite suggestion”.
35. Cf. Laury (2012, p. 234) who – in describing if-conditionals used as directives in Finnish –
points out: “In fact, it appears that indirectness of form may be connected with the contingencies
involved in requests in complex ways, and that there may be quite a bit of cross-linguistic varia-
tion as to which types of forms index requests viewed as problematic in some ways.” In German,
the use of the subjunctive, and thus its hypothetical semantic value, may actually contribute to
making the request less direct.
(11) STRICT MOTHER (REALITY-TV-SHOW: bb #17)36

228 Sa: <all> un_am nächsten tach hat_sie irgendwie,>
an_at the next day have-prs.3sg she somehow
<all> and the next day she somehow,>
229 die (.) EINkaufstÜte (so),
the grocery_bag (so)
the (.) grocery bag (so),
230 die hat das nIch geMERKT,
she [det-art.nom.sg.F] have-prs.3sg this not notice-ptcp
she hadn’t noticed it,
231 <<all> und dann hat- (-)
and then have-prs.3sg
<<all> and then- (-)
232 sie dann das ge[SEHN] ne,>
she then it see-ptcp ptcl
she saw it [then] eh,>
233 Jür: [ja. ]
yeah
[yeah.]
234 Sa: und dann HAT_se dann-
and then have-prs.3sg_she then
and then she then-
235 (weißt halt )(.) da war der ganze verband alles voll[BLU:T. ]
(know-prs.2sg
ptcl) there be-pst.3sg the whole bandage everything
full blood
(you know ) (.) the whole bandage was full
of [blood.]
236 Jür: [<hm]
237 (0.5)
238 Sa: und dann HATse,
and then have-prs.3sg_she
and then she,
239 <<flüsternd> wenn du das noch mal MAchst.> (--)
if you this again ptcl make-prs.2sg
<<whispering> if you ever do that again.> (--)
240 < meine mutter war STRENG.>
my mother be-pst.3sg strict
< my mother was strict.>
241 Jrg: ja aber das ist doch n=!UN!fall.>
yeah but that be-prs.3sg ptcl an accident
yeah but that was just an accident.
The reported threat <<flüsternd > wenn du das noch mal mAchst.> ‘<<whisper-
ing> if you ever do that again.>’ (see l. 239) made by Sabrina’s mother leaves open
36. Cf. also Günthner (2017).

the question of what action she might take, if her daughter were to do something
like that again. The reconstructed wenn-clause forms a prosodic unit of its own
with a nucleus accent and utterance final intonation. One could argue that the
mention of the mother’s threat, in combination with contextualization cues
such as marked prosody, reference to a future action of the addressee, the use
of particular modal adverbs and particles, etc., is sufficient for the daughter to
understand the seriousness of her mother’s warning; so that the articulation of any
possible consequence may be omitted.
Such stand-alone wenn-clauses uttered by a speaker (here the quoted mother),
who has no intention of continuing so deliberately leaves the possible consequence
‘in the air’, reveal striking parallels to the rhetorical device of the “aposiopeses”
(Auer 2007, p. 105; Imo 2014, p. 163–164), which “serve as fully functional ut-
terances in a given context and are intended for the recipients to react to” (Imo,
2014, p. 163). Similar to requests for actions, the formulating of a warning or
threat represents a form of “interpersonal coercion”, and thus, they belong to
those activities, which are susceptible to insubordination (Evans, 2007, p. 392).
Like requests for actions, warnings and threats form “dispreferred social activities”
which – according to Schegloff (1996, p. 448) – may motivate “‘tactful’ trade-offs”.
One way of tactfully dealing with these sensitive activities is implementing them
in form of stand-alone wenn-constructions and thus leaving the contextually
available consequences ‘in the air’.
2.5.3 Exclamations and expressions of stances and assessments:

Speakers in my data, however, also use stand-alone wenn-constructions as
resources for things other than forms of interpersonal coercion (e.g. making
requests, wishes, threats, warnings, etc.). They apply them as a constructional
resource with which to express exclamations and to convey affectionally loaded
assessments and stances.
Excerpt (12) from a reality-TV series, the inhabitants are backbiting about
Marion, a co-inhabitant, who needed pain killers after she had shaved her legs:
(12) SHAVING HER LEGS (REALITY-TV-SHOW: bb #45)
324 Dan: =ich GLAUB,
I believe-prs.1sg
=I think,
325 d_MArion interessIert des net bsonders;
det_marion interest-prs.3sg this not particularly
that marion isn’t really interested in it;
326 Ste: nja außerDEM- (.)
ptcl besides
well besides- (.)
327 wenn man sich SO die BEIne rasIErt.

if one oneself-refl so the legs shave-prs.3sg
if you shave your legs the way she does.
328 (--)
329 Hka: hh° AH°°
330 Ste: oh wir sin Echt FIES ne?
ptcl we be-prs.1pl really mean ptcl
oh we are being really mean aren’t we?

331 [fast (.) geMEIN] schon;
nearly bitchy already
[nearly (.)bitchy] already;
332 Hka: [nö wieSO denn.]
no why ptcl
[no why would you think that.]
The speaker (Steffi) uses a stand-alone wenn-clause nja außerDEM- (.) wenn man
sich SO die BEne rasIErt, ‘well besides- (.) if you shave your legs the way she does,’
(see l. 326–327) to express her indignation about the way Marion shaves her legs.
The prosodic design – the accentuation on both syllables BEI and SIERT with
the marked rise-fall contour on BEI and the oscillating volume – contextualizes
heightened emotive involvement. In leaving out the possible consequence, the re-
cipients are invited to reconstruct the speaker’s stance towards Marion’s behavior.
Here, the prosodic cues (marking indignation),37 the preceding context as well as
shared knowledge about Marion help to interpret speaker’s stance about Marion’s
behavior (i.e., to shave her legs the way she does) as deviant.
Even though, from the point of view of normative grammar, this wenn-clause
is ‘incomplete’ and invites the recipients to reconstruct whatever has been left un-
said, it is treated as an accountable activity by co-participants and becomes a turn
of its own. Prosody fortifies its ‘independence’: the wenn-clause is produced as a
prosodic unit of its own with utterance final intonation. Hanka’s heavy breathing
hh° AH° (see l. 329) underlines her own indignation.
Laury et al. (2013, p. 261) present a similar argument in their analysis of in-
dependent conditionals in Swedish and Finnish: “In such uses, these conditional
clauses are readily interpretable as requests, proposals and suggestions. We have
suggested that these clauses are not inherently subordinate and projecting, but rath-
er they are functionally independent clauses marked with a conditional particle.”
This leads to the following questions:
37. Cf. Christmann & Günthner (1999: 1–33) on contextualizing indignation in everyday
interactions.
a. How can a wenn-clause be interpreted as an independent unit when it is

marked as subordinate and according to our shared grammatical understanding
incomplete?
As the data show, in combination with specific prosodic, grammatical and
semantic features and associated with social action types and local contingen-
cies, the projective force of the initial wenn-clause may be suspended. In these
cases, co-participants treat the single wenn-clause as an autonomous unit, as a
self-contained activity. This supports the idea that syntactical indications do not
provide sufficient or stable cues with which to indicate whether a clause is to be
continued. Participants seem to rely on various other cues – such as prosody, mo-
dality markers (modal particles, subjunctive mood), the type of action indicated
and interactional experience with routinized constructions – in order to interpret
a wenn-clause as the initial part of a compound clause-combining gestalt or as an
independent unit performing a fully-fledged activity.
This issue touches on the topic of “action ascription” (Levinson, 2013) and
the question of how participants assign a particular action – such as a request, a
threat, an assessment, etc. – to a stand-alone wenn-unit. In the data, we can detect
a variety of stand-alone constructions. These range from framing devices to more
or less routinized stand-alone wenn-constructions used to express stances, wishes,
requests, threats etc.38
b. How can we account for these uses of wenn-clauses as independent units,
and thus for the reanalysis of an erstwhile subordinate clause as an independent
one?
Evans (2007, p. 370) argues that insubordinate clauses – “having diachronic
origins as a subordinate clause” – over time have become “reanalysed as standard
constructions”. Such a reanalysis, however, challenges the postulated unidirec-
tionality in grammaticalization (Lehmann, 1982/95 13), which leads from a “less
grammatical to a more grammatical unit”. Grammaticalization theory assumes that
parts of a constructional schema come to have stronger “internal dependencies”
(Haspelmath, 2004, p. 26). Hence, stand-alone wenn-clauses represent counterex-
amples to this unidirectional tendency going from “less to more grammatical” and
thus from “main clause to subordinate clause”, revealing an increase in syntactic
as well as pragmatic autonomy (Günthner 2015a, b, 2017; Hilpert 2015). This
leads to further questions which can only be answered by continuing empirical
research: Under what conditions do we actually find subordinate clauses gaining
main clause status? And how are these conditions related to specific interactive
functions in discourse?
38. Cf. also Günthner (2012), Günthner i. pr. for wenn-fragments with the wenn-unit consisting
of a single wenn-item (e.g., wenn dann morgen Abend. ‘if, then tomorrow evening.’).
3. Conclusion
Speakers of everyday German make use of a broad range of pre-positioned

wenn-constructions, which do neither represent fixed biclausal sentences nor
fixed form-function entities, instead they are rather “openly extendable and have
fuzzy and negotiable boundaries and areas of overlap with other, structurally or
functionally similar, constructions.” (Hopper, 2004, p. 1). However, in spite of this
flexibility, prepositioned wenn-constructions – with the wenn-unit projecting a
following unit (the consequence) – are part of participants’ shared constructional
knowledge, which allows them to function as organizing devices for the dynamics
of social interaction and are implemented as practices for a range of activities:
a. Whereas syntactically integrated wenn-clauses are used to mark closely con-
nected propositions, non-integrated ones tend to be used in contexts in which
the initial wenn-part works as a preparatory device, setting the scene for the
following communicative action. The syntactic architecture of the clause-
combining, thus, iconically reflects on this framing function.
b. Jointly produced wenn-gestalts not only demonstrate how participants in
the emergence of talk-in-interaction co-construct this grammatical gestalt,
but they also provide ‘live’ evidence that participants attend to commonly
shared syntactic gestalts. Even though joint wenn-constructions exemplify
how participants co-align in instantiating complex clausal structures, second
speakers may use syntactic continuations to express discordant and disaffilia-
tive stances.
c. Participants in everyday interaction exploit the projective force of initial
wenn-units to introduce complex segments which extend over several turn
construction units and thus ensure that the speaker has the opportunity to
finish her/his multi-unit turn. This turn-holding function is used for various
types of ‘big packaging’, e.g., complex (argumentative) sequences which extend
over several turn construction units.
d. Speakers, however, not only produce wenn-units followed by a consequence,
but they also implement non-projecting, stand-alone wenn-units which
constitute self-contained communicative actions. These formats are used as
more or less routinized practices for implementing wishes, requests, warnings
and threats and thus, social activities which perform “interpersonal coercion”
(Evans, 2007, p. 392). In addition stand-alone wenn-constructions can also be
used for affectively loaded exclamations, emotional stances and assessments.
Wenn-constructions in everyday interactions not only show multifaceted gestalts,
but participants use these formats for implementing a range of activities. Even
though the use of grammatical constructions is based on intersubjectively shared
knowledge (which in turn makes projections, co-constructions etc. possible),

wenn-constructions do not represent a stable biclausal pattern. Instead of assum-
ing fixed form-meaning pairs as mental conceptualizations that are then reactual-
ized in situated discourse, it seems more appropriate to treat wenn-constructions
as patterns of orientation, whose instantiations occur in the ‘hic et nunc’ of the
ongoing interaction, responding to local contingencies and other interactional
factors (Günthner 2007, 2011a; Breyer, Ehmer & Pfänder 2011).
Literature
Auer, P. (1996). The pre-front field position in spoken German and its relevance as a gram-
maticalization position. Pragmatics 6(3), 295–322. https://doi.org/10.1075/prag.6.3.03aue
Auer, P. (2000). Pre- and post-positioning of wenn-clauses in spoken and written German. In E.
Couper-Kuhlen & B. Kortmann (Eds.), Cause, condition, concession, contrast: cognitive and
discourse perspectives (pp. 173–204). Berlin/Boston: de Gruyter.
https://doi.org/10.1515/9783110219043.2.173
Auer, P. (2005). Projection in interaction and projection in grammar. Text 25, 7–36.
Auer, P. (2007). Syntax als Prozess. In H. Hausendorf (Ed.), Gespräch als Prozess. Linguistische
Aspekte der Zeitlichkeit verbaler Interaktion. (pp. 95–124). Tübingen: Narr.
Auer, P. (2009a). On-line syntax: Thoughts on the temporality of spoken language. Language
Auer, P. (2009b). Projection and minimalistic syntax in interaction. Discourse Processes 46(2),
180–205. https://doi.org/10.1080/01638530902728934
Auer, P. & Pfänder, S. (2011). Constructions: Emergent or emerging? In P. Auer & S. Pfänder
(Eds.), Constructions: Emerging and Emergent (pp. 1–12). Berlin/Boston: De Gruyter.
https://doi.org/10.1515/9783110229080.1
Auer, P. & Lindström, J. (2015). Left/right asymmetries and the grammar of pre- vs. postposi-
tioning in German and Swedish talk-in-interaction. InLiSt 56.
Bergmann, J. (1992). Konversationsanalyse. In Flick, U. et al. (Eds.), Handbuch qualitativer
Sozialforschung (pp. 213–218). München: Psychologie Verlags Union.
Boogaart, R. & Verheij, K. (2013). Als dát geen insubordinatie is! De pragmatiek van zelfstandige
conditionele zinnen. In T. Janssen, & J. Noordegraaf (Eds.), Honderd jaar taalwetenschap
(pp. 12–28). Amsterdam/Münster: Nodus Publikationen.
Breyer, T., Ehmer, O. & Pfänder, S. (2011). Improvisation, temporality and emergent construc-
tions. In P. Auer & S. Pfänder (Eds.), Constructions: Emerging and Emergent (pp. 186–217).
Berlin/Boston: De Gruyter. https://doi.org/10.1515/9783110229080.186
Brown, P. & Levinson, S. C. (1978/87). Politeness. Some universals in language usage. Cambridge,
MA: Cambridge University Press.
Brugmann, K. (1918): Verschiedenheit der Satzgestaltung nach Maßgabe der seelischen Grund-
funktionen in den indogermanischen Sprachen. Leipzig: B.G. Teubner.
Christmann, G. & Günthner, S. (1996). Sprache und Affekt. Die Inszenierung von Entrüstung
im Gespräch. Deutsche Sprache 1, 1–33.
Couper-Kuhlen, E. & Barth-Weingarten, D. (2011). A system for transcribing talk-in-interac-

tion: GAT 2. Gesprächsforschung – Online-Zeitschrift zur verbalen Interaktion 12, 1–51.
Couper-Kuhlen, E., Fox, B. & Thompson, S. (2014). Forms of responsitivity: Grammatical
formats for responding to two types of request in conversation. In S. Günthner, W. Imo
& J. Bücker (Eds.). Grammar and Dialogism. Sequential, Syntactic, and Prosodic Patterns
between Emergence and Sedimentation (pp. 109–138). Berlin/Boston: de Gruyter.
Dancygier, B. & Sweetser, E. (2000). Constructions with if, since, and because: Causality,
epistemic stance, and clause order. In E. Couper-Kuhlen & Kortmann, B. (Eds.), Cause,
Condition, Concession, Contrast: Cognitive and Discourse Perspectives (p. 111–142). Berlin/
Boston: de Gruyter. https://doi.org/10.1515/9783110219043.2.111
Deppermann, A. (1999). Gespräche analysieren. Eine Einführung in konversationsanalytische
Methoden. Opladen: Leske & Budrich.
Deppermann, A. & Günthner, S. (Eds.) (2015). Temporality in Interaction. Amsterdam: John
Benjamins. https://doi.org/10.1075/slsi.27
Du Bois, J. W. (2014). Towards a dialogic syntax. Cognitive Linguistics 25(3), 359–410.
https://doi.org/10.1515/cog-2014-0024
Evans, N. (2007). Insubordination and its uses. In I. Nikolaeva (Ed.), Finitness: theoretical and
empirical foundations (pp. 366–431). Oxford: Oxford University Press.
Fauconnier, G. (1985): Mental Spaces. Cambridge, MA: MIT Press.
Ford, C. (1993). Grammar in Interaction. Cambridge, MA: Cambridge University Press.
https://doi.org/10.1017/CBO9780511554278
Greenberg, J. H. (1963). Some universals of grammar with particular reference to the order
of meaningful elements. In J. H. Greenberg (Ed.), Universals of Language (pp. 73–113).
Cambridge, Mass.: MIT Press.
Günthner, S. (1999). Wenn-Sätze im Vor-Vorfeld: Ihre Formen und Funktionen in der gespro-
chenen Sprache. Deutsche Sprache 3, 209–235.
Günthner, S. (2006). Was ihn trieb, war vor allem Wanderslust: Pseudocleft-Konstruktionen im
Deutschen. In S. Günthner, & W. Imo (Eds.), Konstruktionen in der Interaktion (pp. 59–90).
Berlin/Boston: de Gruyter. https://doi.org/10.1515/9783110894158.59
Günthner, S. (2007). Brauchen wir eine Theorie der gesprochenen Sprache? Und: wie kann sie
aussehen? GIDI-Arbeitspapierreihe (Grammatik in der Interaktion). http://arbeitspapiere.
sprache-interaktion.de/2007/06/ Last access on 24.09.18.
Günthner, S. (2008a). ‘Die Sache ist …‘: eine Projektorkonstruktion im gesprochenen Deutsch.
Zeitschrift für Sprachwissenschaft 27(1), 39–72. https://doi.org/10.1515/ZFSW.2008.003
Günthner, S. (2008b). Projektorkonstruktionen im Gespräch: Pseudoclefts, die Sache ist-
Konstruktionen und Extrapositionen mit es. Gesprächsforschung 9, 86–114.
Günthner, S. (2009). Konstruktionen in der kommunikativen Praxis. Zur Notwendigkeit einer
interaktionalen Anreicherung konstruktionsgrammatischer Ansätze. Zeitschrift für ger-
manistische Linguistik 37(3), 402–426. https://doi.org/10.1515/ZGL.2009.029
Günthner, S. (2011a). Between emergence and sedimentation. Projecting constructions in Ger-
man interactions. In P. Auer & S. Pfänder (Eds.), Constructions: Emerging and Emergent
(pp. 156–185). Berlin/ Boston: de Gruyter. https://doi.org/10.1515/9783110229080.156
Günthner, S. (2011b). N be that-constructions in everyday German conversation. A renanalysis
of ‘die Sache ist/das Ding ist’ (‘the thing is’)-clauses as projector phrases. In R. Laury, & R.
Suzuki (Eds.), Subordination in Conversation (pp. 11–36). Amsterdam: John Benjamins.
Günthner, S. (2012). Die Schriftsprache als Leitvarietät die gesprochene Sprache als Abweic-
hung? ‘Normwidrige’ wenn-Sätze im Gebrauch. In S. Günthner et al. (Eds.), Kommunika-
tion und Öffentlichkeit (pp. 61–84). Berlin/Boston: de Gruyter.
https://doi.org/10.1515/9783110289916.61
Günthner, S. (2014). The dynamics of dass-constructions in everyday German interactions – a
dialogical perspective. In S. Günthner, W. Imo & J. Bücker (Eds.), Grammar and Dialogism.
Sequential, syntactic, and prosodic patterns between emergence and sedimentation (pp.
179–206). Berlin/Boston: de Gruyter.
Günthner, S. (2014). This is a paper published in the following volumn: S. Günthner, W. Imo
& J. Bücker (Eds.), Grammar and Dialogism. Sequential, syntactic, and prosodic patterns
between emergence and sedimentation (pp. 179–206). Berlin/Boston: de Gruyter.
Günthner, S. (2015a). ‚Geteilte Syntax‘: Kollaborativ erzeugte dass-Konstruktionen. In A. Ziem
& A. Lasch (Eds.), Konstruktionsgrammatik IV. Konstruktionen als soziale Konventionen
und kognitive Routinen (pp. 25–40). Tübingen: Stauffenburg.
Günthner, S. (2015b). Ko-Konstruktionen im Gespräch: Zwischen Kollaboration und Konfron-
tation. In U. Dausendschön-Gay, E. Gülich & U. Krafft (Eds.), Ko-Konstruktionen in der
Interaktion. (pp. 55–74). Bielefeld: transcript Verlag.
https://doi.org/10.14361/9783839432952-005
Günthner, S. (2015c). A temporally oriented perspective on connectors in interaction: und zwar
(,namely/in fact‘)-constructions in everyday German conversations. In A. Deppermann &
S. Günthner (Eds.), Temporality in Interaction (pp. 237–266). Amsterdam: John Benjamin.
Günthner, S. (2017). Alleinstehende Nebensätze: Insubordinierte wenn-Konstruktionen in der
kommunikativen Praxis. In Y. Ekinci, E. Montanari & L. Selmani (Eds.), Grammatik und
Variation. (pp. 97–111). Heidelberg: Synchron Wissenschaftsverlag der Autoren.
Günthner, S. & Hopper, P. (2010). Zeitlichkeit & sprachliche Strukturen: Pseudoclefts im Eng-
lischen und Deutschen. Gesprächsforschung 11, 1–28.
Günthner, S. & Imo, W. (Eds.) (2006). Konstruktionen in der Interaktion. Berlin/New York: de
Gruyter. https://doi.org/10.1515/9783110894158
Günthner, S., Imo, W. & Bücker, J. (Eds.) (2014). Grammar and Dialogism. Sequential, syntactic,
and prosodic patterns between emergence and sedimentation. Berlin/Boston: de Gruyter.
https://doi.org/10.1515/9783110358612.1
Günthner, S. (i. pr.). Wenn-Konstruktionen im Gespräch. Zur Verwobenheit kognitiver und
interaktionaler Faktoren bei der Realisierung grammatischer Muster. In A. Binanzer, J.
Gamper & V. Wecker (Eds.), Prototypen - Schemata - Konstruktionen. Untersuchungen zur
deutschen Morphologie und Syntax. Berlin: de Gruyter.
Haspelmath, M. (2004). Coordinating constructions: an overview. In: Haspelmath, M. (Ed.),
Coordinating constructions (pp. 3–39). Amsterdam: Benjamins.
https://doi.org/10.1075/tsl.58.03has
Hilpert, M. (2015). Kollaborative Insubordination in gesprochenem Englisch: Konstruktion
oder Umgang mit Konstruktionen? In A. Ziem & A. Lasch (Eds.), Konstruktionsgrammatik
IV. Konstruktionen als soziale Konventionen und kognitive Routinen (pp. 25–40). Tübingen:
Stauffenburg.
Hopper, P. (1987). Emergent Grammar. In Berkeley Linguistic Society (Ed.), General Session and
Parasession on Grammar and Cognition (pp. 139–157). Berkeley CA: Berkeley Linguistics
Society.
Hopper, P. (1988). Emergent Grammar and the A Priori Grammar Postulate. In D. Tannen (Ed.),
Linguistics in Context (pp. 117–133). Norwood: Ablex.
Hopper, P. (2004). The openness of grammatical constructions. Chicago Linguistic Society 40,
153–175.
Hopper, P. (2011). Emergent grammar and temporality in interactional linguistics. In P. Auer
& S. Pfänder (Eds.), Constructions: Emerging and Emergent (pp. 22–44). Berlin/ Boston: de
Gruyter. https://doi.org/10.1515/9783110229080.22
Hopper, P. & Thompson, S. A. (2008). Projectability and clause combining in interaction. In R.
Laury (Ed.). Crosslinguistic Studies of Clause Combining. The multifunctionality of conjunc-
Imo, W. (2013). Sprache in Interaktion. Analysemethoden und Untersuchungsfelder. Berlin: de
Gruyter. https://doi.org/10.1515/9783110306323
Imo, W. (2014). Elliptical structures as dialogical resources for the management of understand-
ing. In S. Günthner, W. Imo & J. Bücker (Eds.), Grammar and Dialogism. Sequential, Syn-
tactic, and Prosodic Patterns between Emergence and Sedimentation (pp. 139–176). Berlin/
Boston: de Gruyter.
Jespersen, O. (1924). The Philosophy of Grammar. London: George Allen, and Unwin.
König, E. & Van der Auwera, J. (1988). Clause integration in German and Dutch conditionals,
concessive conditionals, and concessives. In Haiman, J. & S. A. Thompson (Eds.), Clause
Combining in Grammar and Discourse (pp. 101–134). Amsterdam: John Benjamins.
https://doi.org/10.1075/tsl.18.07kon
Laury, R. (2012). Syntactically Non-Integrated Finnish jos ‘If ’-Conditional Clauses as Direc-
tives. Discourse Processes 49, 213–242. https://doi.org/10.1080/0163853X.2012.664758
Laury, R., Lindholm, C. & Lindström, J. (2013). Syntactically non-integrated conditional clauses
in spoken Finish and Swedish. In E. Havu & I. Hyvärinen (Eds.), Comparing and contrasting
syntactic structures. (pp. 231–269). Helsinki: Société Néophilologique de Helsinki.
Lehmann, C. (1982/95). Thoughts on Grammaticalization. München/Newcastle: Lincom.
Lerner, G. H. (1991). On the syntax of sentences-in-progress. Language in Society 20, 441–458.
https://doi.org/10.1017/S0047404500016572
Lerner, G. H. (2002). Collaborative turn sequences. In G. H. Lerner (Ed.) Conversation Analysis.
Studies from the first generation (pp. 225–256). Amsterdam: Benjamins.
Leipzig Glossing Rules (2015). The Leipzig Glossing Rules: Conventions for interlinear mor-
pheme-by-morpheme glosses. URL: https://www.eva.mpg.de/lingua/pdf/Glossing-Rules.
pdf. Last access on 24.09.18.
Levinson, S. C. (2006). Cognition at the heart of human interaction. Discourse Studies 8, 85–93.
https://doi.org/10.1177/1461445606059557
Levinson, S. C. (2013). Action Formation and Ascription. In J. Sidnell & T. Stivers (Eds.), The
Handbook of Conversation Analysis (pp. 103–130). Chichester: Blackwell.
Linell, P. (2009). Rethinking Language, Mind, and World Dialogically: Interactional and contex-
tual Theories of Human Sense-Making. Charlotte, NC: Information Age Publishing.
Luckmann, T. (1990). Social communication, dialogue and conversation. In I. Marková & K.
Foppa (Eds.), The Dynamics of Dialogue (pp. 45–61). New York: Harvester Wheatsheaf.
Maschler, Y. (2009). Metalanguage in Interaction: Hebrew Discourse Markers. Amsterdam: John
Benjamins. https://doi.org/10.1075/pbns.181
Matthiessen, C. & Thompson, S. A. (1988). The Structure of Discourse and ‘Subordination’. In
J. Haiman & S. A. Thompson (Eds.): Clause Combining in Grammar and Discourse (pp.
275–330). Amsterdam: John Benjamins. https://doi.org/10.1075/tsl.18.12mat
Metschkowa-Atanassova, Z. (1983). Temporale und konditionale ‘wenn’-Sätze. Düsseldorf:
Pädagogischer Verlag Schwann.
Oppenrieder, W. (1989). Selbständige Verb-Letzt-Sätze: Ihr Platz im Satzmodussystem und ihre

intonatorische Kennzeichnung. In H. Altmann, A. Batliner & W. Oppenrieder (Eds.), Zur
Intonation von Modus und Fokus im Deutschen (pp. 163–244). Tübingen: Niemeyer.
structions in French talk-in-interaction. In R. Laury & R. Suzuki (Eds.), Subordination in
Conversation. A cross-linguistic perspective (pp. 103–148). Amsterdam: John Benjamins.
Schegloff, E. A. (1984). On some questions and ambiguities in conversation. In J. M. Atkinson &
J. Heritage (Eds.), Structures of Social Action. Studies in Conversation Analysis (pp. 28–52).
Cambridge MA.: Cambridge University Press.
Schegloff, E. A. (1996). Some Practices for Referring to Persons in Talk-in-Interaction: A Partial
Sketch of a Systematics. In B. Fox (Ed.), Studies in Anaphora (pp. 437–485). Amsterdam:
John Benjamins. https://doi.org/10.1075/tsl.33.14sch
Schütz, A. (1932/1973). Der sinnhafte Aufbau der sozialen Welt: Eine Einleitung in die verstehende
Soziologie. Wien: Julius Springer. https://doi.org/10.1007/978-3-7091-3108-4
Selting, M. et al. (2009). Gesprächsanalytisches Transkriptionssystem 2 (GAT 2). Gesprächs-
forschung 10, 353–402.
Silverstein, M. (1993). Metapragmatic discourse and metapragmatic function. In J. Lucy (Ed.),
Reflexive Language: Reported Speech and Metapragmatics (pp. 33–58). Cambridge MA:
Cambridge University Press. https://doi.org/10.1017/CBO9780511621031.004
Stirling, L. (1999). Isolated if-clauses in Australian English. In P. Collins & D. Lee (Eds.), The
Clause in English. In honour of Rodney Huddleston. (pp. 273–294). Amsterdam: John Ben-
jamins. https://doi.org/10.1075/slcs.45.18sti
Sweetser, E. E. (1990). From Etymology to Pragmatics. Metaphorical and cultural aspects of
semantic structure. Cambridge: University Press.
https://doi.org/10.1017/CBO9780511620904
Vallauri, E. L. (2004). Grammaticalization of Syntactic Incompleteness: Free Conditionals in
Italian and Other Languages. SKY/Journal of Linguistics 17, 189–215.
Wegner, L. (2010). Unverbundene WENN-Sätze in der gesprochenen Sprache. GIDI-Arbeitspa-
pierreihe. URL: http://centrum.sprache-interaktion.de/cesi-umfasst/dfg-projekt-gramma-
tik-in-der-interaktion-gidi/arbeitspapiere/arbeitspapier28.pdf. Last access on 24.09.18.
Zifonun, G. et al. (1997). Grammatik der deutschen Sprache. Band 1–3. Berlin/New York: de
Gruyter.
Chapter 8
Grammatical coordination of
embodied action
The Estonian ja ‘and’ as a temporal organizer of
Pilates moves
Leelo Keevallik
This paper looks at the Estonian coordinating conjunction ja ‘and’ in video-

recorded Pilates classes, focusing on the instructors’ practical problem of making
the students perform proper movement sequences. It shows how grammatical
coordination emerges within a multimodal activity in which the instructor’s talk
both directs and responds to student performance. As opposed to the frequent
juxtaposition of clauses without connectors, explicit coordination with ja is
used for the overall structuring of the class as well as the temporal extension
of talk to achieve synchronicity of vocal and embodied behavior. In contrast to
formal theories that consider grammar as a device for coherent expression of
pre-planned propositions, this study argues that grammatical structure emerges
as part of practical action across participants and modalities.
Keywords: coordination, conjunction, Estonian, synchronization
1. Introduction
This study targets syntactic coordination in Estonian Pilates classes. Coordination

in grammar has been defined as the “morphosyntactic means of linking two
clauses of equal grammatical status” (Payne 1997: 336, Haspelmath 2007). The
fact that the two clauses are grammatically coordinated implies that (1) the two
clauses have more or less the same function in terms of the event structure of the
text, and (2) they are presented as being conceptually linked in some way (Payne
1997: 337). Items such as and in English, and ja ‘and’ in both Estonian and Finnish
have been considered prototypical or neutral coordinating conjunctions (Erelt.
et al. 1993: 216, Hakulinen et al. 2004: 1029, 1040), which may even lack their
https://doi.org/10.1075/slsi.32.08kee
222 Leelo Keevallik
own meaning (Kalliokoski 1989: 142–143; Schiffrin 1987: 128–150). On the basis
of conversational data in English the main function of and has been claimed to be
something very generic: the coordination of idea units and continuing speaker’s
action, as a link at local as well as global levels of discourse (Schiffrin 1987: 128,
131, 141). Similarly, the descriptive dictionary of Estonian registers the fact that
ja not only coordinates clauses but can also tie the upcoming sentence to a larger
context (EKSS). In comparison, the Dutch en ‘and’ has shown to mark continua-
tion across turns (Mazeland & Huiskes 2001). Perhaps surprisingly, Turk (2004)
has instead argued that and functions as a “discontinuity marker” in everyday
conversation, as it may achieve a return to a prior matter and thereby smooth over
certain discontinuities. In more specific medical and aviation contexts, it has been
shown how the English and-preface frames questions as the next thing on the
agenda thus structuring larger activities (Heritage & Sorjonen 1994, Nevile 2006).
The different pronunciation variants in English have been shown to correlate with
the syntactic-pragmatic scope (Barth-Weingarten 2012) and in combination with
uhm, and it has been shown to tie back to earlier talk (Local 2004). In short, prior
research has pointed out that the generic coordinating conjunction is basically
void of meaning outside specific contexts, and that it may be involved in achieving
coherence across larger segments of text and talk.
In specific contexts, however, the generally additive Estonian or Finnish ja
can be interpreted as temporal, contrastive, or causal (Kalliokoski 1989, Erelt et al.
1993: 278–280, Hakulinen et al. 2004: 1040). The Estonian comprehensive gram-
mar furthermore states that the additive conjunction ja can be used for marking
simultaneity or listing of events (Erelt et al. 1993: 278–280). In Finnish the listing
capacity of ja ‘and’ has been thoroughly analyzed in turn-endings, where a ja marks
the list as non-exhaustive and urges recipient inference (Koivisto 2011: 86–126).
Especially the listing pattern suggests that ja is usable for marking parallel (syntac-
tic or phrasal) structures, as in Nad tulid koju ja hakkasid sööma ‘They came home
and started to eat’. This parallelism also figures in the verbal guidance of repetitive
Pilates exercises, albeit in terms of action. Moreover, Pilates practice happens in
real time and thus also reflects the temporal meaning of ja. In earlier linguistic
accounts, temporality has been partially ascribed to the general iconicity of clause
arrangement: what is verbalized first happens first, for example also in conditional
clauses where the condition comes first (e.g. Kalliokoski (1989: 233) on Finnish).
Parallelism and temporality are essential for the activity of Pilates practice and the
present study looks at whether and how ja contributes to achieving both features.
The paper targets an essential practical problem for Pilates instructors – how to
make students achieve proper movement sequences (see Mondada (2018) on simi-
lar issues on a race circuit). Progressivity here concerns the exercises themselves
as well as transitions between them: some physical moves have to be implemented
Chapter 8. Grammatical coordination of embodied action 223
as a preparation before others can be performed, whereas other sequences are a

part of the exercise itself, and all of them have to be accomplished in temporal
succession. The instructor’s utterances are often designedly synchronous with the
students’ moves and thus indexically mark temporalities in the class. The utter-
ances may be clausal, phrasal, consist of a single lexical word or even a mere non-
lexical vocalization. It seems, however, that at particular moments more distinct
marking of succession is necessary and it is at these moments the coordinating
conjunction ja ‘and’ is deployed. Nevertheless, ja is but one option of coordination,
while in the so-called “zero strategy” the phrases or clauses are simply juxtaposed
(Payne 1997: 337), perhaps tied together via prosodic means in spoken language.
In Pilates classes the role of various stylized patterns of projective prosody is still to
be worked out, but overall a continuation without explicit connectors is a default
in this activity. In order to carve out the function of ja the current paper looks
contrastively at the two options. It shows how grammatical coordination emerges
within a multimodal interactional activity, where instructor talk both directs and
responds to student performance, often during a single word, such as ja. Words
and sounds can be lengthened and produced with characteristic intonation in
order to reinforce the progress of the exercise as well as to iconically represent its
appropriate length or the relevant amount of energy. The coordination of clauses
thereby emerges not simply as a result of a cognitive processing of propositions but
as a reflection and accomplishment of a real-time embodied event.
2. The data
The data come from a video-recording of four Pilates classes in Estonian (total
duration about 4 hours, with 5–12 participants per class). Pilates is a fitness system
that focuses on strength and stamina and aims at an economic flow of movement,
including elegant transitions from one exercise to the next. It makes essential
use of breathing, which can be seen in the excerpts below. The classes are offered
to the general public. The students had varying experience of the training style
but none of them seemed a total beginner. The classes were recorded with 1 to 3
cameras at different angles and the participants agreed to the use of the record-
ings for research purposes. There was no music playing, as is otherwise often the
case in, e.g., yoga classes, and this makes the activity setting especially interesting
for linguistic research. The exercise rhythms are created here entirely based on
the interplay between the instructor’s voice and the moving bodies of both the
instructor and the students.
The study is based on a collection of instances where the instructor uses the
word ja ‘and’ as a conjunction before clauses. This is a frequent word in spoken
224 Leelo Keevallik
Estonian, ranking as the second most used word in everyday interaction (3% of the
overall word amount of 52 000) and the third in institutional interaction (2% out
of 38 000) in the Tartu corpus (Hennoste et al. 2000: 275). Ja ‘and’ is also extremely
frequent in the current data, much more so than the functionally similarly generic
Estonian conjunction et ‘that/so/and’ (Keevallik 2000) that seems to be almost as
frequent as ja in talk-focused activities. For example, in the above brief overview
(Hennoste et al. 2000: 275) et covered 2% of lexical items both in everyday and
institutional interaction. In sharp contrast to these results on conversation, ja
occurred 105 times before clausal structures during 30 minutes of a Pilates class,
while et did so only 6 times. Furthermore, another coordinating conjunction, the
adversative aga ‘but’ only occurred 11 times in the data. The counts, even though
not an aim in themselves in the current study, reflect the distinct status of ja in the
activity context of a Pilates class. We will thus be focusing on a linguistic device
that is obviously central to this activity type. In comparison, the production of a
next clause without a connector occurred 383 times during the same time and thus
seems to be the default strategy, against which to study the specific accomplish-
ment of the structures coordinated with ja.
The collection in this chapter is limited to clausal structures. In addition to
combining clauses, the conjunction ja is frequently deployed before phrasal and
single-word structures in Pilates classes. In this position it often constitutes part of
a so-called formula that is used repetitively and rhythmically during the simulta-
neously repeated physical exercise, such as ja üles – ja alla ‘and up and down’, jalg
ette – ja jalg taha ‘leg to the front and leg to the back’ (Keevallik, in press). Note
that the occurrence of ja in these formulae were not counted among the 105 cases
above, unless the formula involved a grammatically complete clause, such as an
imperative verb (phrase). An example would be ja hinga sisse – ja hinga välja ‘and
breathe in, and breathe out’. There were altogether 26 instances included in the
study where the repetitive formulae contained an imperative.
A further complication of the current collection is the fact that the phonologi-
cally almost identical item ja is used as an affirmative response in Estonian. It is,
however, generally possible to distinguish the ‘yes’ from the ‘and’ on the basis of
their prosody and sequential placement in relation to talk or physical movement,
i.e. the action accomplished. In particular between clauses uttered by the same
speaker, the use of the response particle is not very likely, while it still occurs
as a positive reaction to some student moves. Possible ambiguous cases will be
discussed in detail in Section 6.
3. Coordinating consecutive clauses and moves
The first relatively straightforward Excerpt (1) illustrates the specifics of coordi-
nating clausal structures in a Pilates class. We will focus on lines 4–5: hoia ja tuled
maha ‘hold and you come down’. This is a characteristic indexical instruction,
consisting of two consecutive clauses that are coordinated with ja ‘and’. The clauses
relate to actions that take place successively – first the students have to hold a
position and then come down to the floor. However, in the activity context of a
Pilates class these instructions are synchronized with the students’ moving bodies,
as shown in the figures. The exercise is a so-called plank where the students have to
furthermore slide (or “travel”, as the instructor says (sõida, in line 1)) forward and
back on one foot. While the instructor orders them to lower their leg in line 3, all
but one student manages to do so during the uttering of the word alla ‘down’ (see
Figure 1). This degree of minute synchronization is of course possible because of a
high level of predictability of the moves to be done. As it appears, at least some of
the students are familiar with the exercise and thus able to lower their leg as soon
as the instruction is launched. At the same time, lowering their leg is basically the
only thing that can happen in this quite strenuous position. A moment later, while
the instructor instructs hoia ‘hold’ everybody is in plank (as shown in Figure 2)
but as soon as they hear the first sound of the coordinating ja ‘and’ in line 5, they
start bending their legs and arriving on the floor. As can be seen in Figure 3, two
students have already landed in the middle of the lengthened ja, and two more are
bending their legs. The last one (of the five students observable in the video) to
lower her legs is the person who is touched by the instructor but even she manages
to lower them before the instruction tuled maha ‘you come down’ is terminated.
The target lines are marked with an arrow.
(1) Hold
01 INS: sõida,
travel:IMP
travel,
02 tagasi,
back,
03 .h all:a?#1
down?
04 hoi#2a?
hold:IMP
hold?
226 Leelo Keevallik
#1 All but one leg down already. #2 Holding plank in synchrony
with the imperative.
05 -> J::A#3:: tuled m:ah#4a:.
and come:2SG down
and you come down.
#3 Two knees on the floor, #4 All knees on the floor.

one person just landing.
This excerpt illustrates the role of ja as a temporal coordination device. On the

one hand, it is used to project a termination instruction to lower the bodies, and
thus arrive at the closure of this difficult exercise. In fact, holding the position,
instructed in line 4, is already an activity-specific means of projecting a closure,
as positions are often held at the end of a series of exercises, but at this moment
they are not to return to the starting position quite yet. On the other hand, the
lengthened ja extends the time before the instruction is actually uttered (also of-
fering an opportunity for the extension of the strenuous position for those who
want to be challenged). As can be seen in Figure 3, four out of five students in the
picture pick up on this cue during the production of the conjunction. One of them
is also gently touched on the back, perhaps individually suggesting a downward
movement via haptic means. In that way the J::A:: both organizes the exercise
but also fine-tunes into the students’ current position, enabling everybody to be
in basic synchrony with the instructor’s voice, which continues without a single
pause throughout this segment of the exercise. The conjunction ja is thus not only
a marker of succession of events formulated in the abstract but also achieved in
real time here and now. It is uttered prosodically as the beginning of an intonation
unit that will then also include the entire upcoming clause.1 It is thus a central
rhythmic coordination device in the activity.
Crucially, ja may carry major stress, as it also does in the current excerpt. In
fact, it is the longest and most prominent word in this entire excerpt. This suggests
that its role in the activity goes beyond mere coordination of successive moves. We
can, for example, notice that it is used at the junction between the strenuous part of
the exercise in the plank position and the return to a resting position on all fours.
As soon as its first sound is uttered, some of the students use this as a sign to return
to resting position. It seems that the clause-combining ja provides them with an
important cue in the trajectory of the exercise and organizes larger trajectories of
the Pilates practice, among other things already projecting a transition to the next
exercise. This organization will be the focus of the ensuing analysis.
4. Clausal units and the organization of the activity: Ja-clauses launching

the termination of an exercise
The instructor talks throughout the class. Not all of her talk, however, consists of
clauses. In order to guide and synchronize the students’ practice, the instructor
regularly uses so-called formulae, i.e. reduced versions of the original full instruc-
tion, which can be single words, phrases, or simply juxtaposed words (Keevallik,
in press). One feature of these repeated formulae is that they can be prefaced by
ja ‘and’, which seems to be a suitable device for creating and maintaining rhythm
(ibid.). Among other things, because of its open syllable structure ja is easy to
lengthen. Excerpt (2) starts with a segment of formula repetition in lines 1–4, with
coherent intonation units on consecutive lines of the transcript. By the beginning
of the excerpt the instructor has reduced the original explanation of the exercise
of a leg lift into a repetitive pattern involving merely three words põlv ‘knee’, lagi
‘ceiling’, and ja ‘and’. She accompanies the students’ leg lifts towards the ceiling
and lowering back to the knee with this repetitive structure, where each segment
is prefaced with a ja, altogether creating a singsong quality of the instruction and
providing the pace of the exercise. The conjunction is furthermore lengthened, so
that the instructor’s speech covers the entire duration of the students’ leg move-
ments downward and upward, respectively.
1. The fact that it initiates an intonation unit at the same time determines its role as a conjunc-
tion rather than an affirmative response here.
228 Leelo Keevallik
(2) And flex

01 INS:j:a:: põl:v?
and knee?
02 ↑ja:: lagi. (0.1)
and ceiling.
03 ja:: põl:v?
and knee?
04 ↑ja: la#1gi.
and ceiling
#1 Legs up in synchrony with #2 Two legs flexing,

the word ‘ceiling’. two more on their way down.
05 tööta r:eie SIseküljega.

work with the inside of the thigh.
06 -> j:a:: kõver#2da? (.)
and flex?
07 -> ja:: kõverda?
and flex?
08 -> ja vaheta jalga.
and switch legs.
09 vaadake et eee >tugijalg ära ei vaju.<
check that the supporting leg doesn’t slack.
10 all:a? (0.2)
down?
11 ↑Üle:s. (.)
up.
12 All:a, (.)
down,
13 ↑ÜLe:s, (.)
up,
14 All:a,
down,
15 -> <& ↑Üle:s &> ja:: kõ#3verda?
up and flex?
#3 Two legs bending in response to ‘and’.

16 hoia? (.)
hold?
17 ja puusad maha.
and hips down.
The formula can also function without ja. This can be seen in lines 10–15, where
only two words, alla ‘down’ and üles ‘up’ are uttered during the very same exercise
with a different leg. Even though these words are also lengthened, like those in the
above ja-prefaced formula, in this case short pauses follow every word to achieve
synchrony with the body movements. This slight difference in the use of silence
can be considered yet another indication that the ja may play a role in extending
the verbal accompaniment, so that the instructor’s speech emerges as well-timed
with the students’ exercising bodies.
While these formulae (ja põlv – ja lagi, alla – üles) consisted of non-clausal ele-
ments, it is also evident in the above excerpt how especially one-word clauses can
be deployed in a similar way. The imperative kõverda ‘flex:IMP’ is used in lines 6,
7, and 15. Even though an imperative constitutes a grammatical clause on its own,
this instruction is nevertheless built indexically on the currently ongoing bodily
movement as well as prior knowledge of the exercise, i.e. which body part should
be flexed and how. Finally, a clause consisting of a verb and an object (vaheta
jalga ‘switch legs’) in line 8 is also prefaced by ja. We thus have a complex clausal
structure in lines 5–8; tööta reie siseküljega ja kõverda ja kõverda ja vaheta jalga
‘work with the inside of the thigh and flex and flex and switch legs’. It emerges in
instructor’s talk in real time across consecutive moves by the students. Part of it is
a reflexive comment on the students’ performance, and an instruction to correct
it during the entire performance of the exercise (line 5) and part of it is a directive
to be immediately complied with. Interestingly, the compliance begins as soon as
the ja in line 6 is uttered and before the imperative directive itself: in figure 2 we
230 Leelo Keevallik
can see two people with their legs almost touching the floor before the impera-
tive ja kõverda ‘and flex’ is complete. Thus, their embodied response emerges in
achieved synchrony with the directive itself. Crucially, had it been just one person
it may have reflected her exhaustion or giving up, but it is clearly two out of six in
Figure 2. The ja in line 6 actually begins at the moment when their legs are at the
highest point, copying the prosodic pattern of lagi ‘ceiling’ in line 4 but then being
lengthened into a downward intonation curve, as if iconically reflecting their legs
having to come down too and waiting for at least some students to be able to move
in synchrony with the ensuing instruction. This kind of a lengthened and prosodi-
cally iconic ja is apparently sufficient for the students to react and initiate a return
to a less strenuous position. We also saw a similar pattern in Excerpt (1). Crucially
for the overall organization of the class, the rhythmic instruction ceases in line 5,
after which there is no return to either the previous rhythm or the formula.
Excerpt (2) illustrates how, in contrast to formulae, the ja-prefaced clauses are
deployed at moments of transition between different exercises. The first formula
ends in line 4 and in line 5 the instructor utters a correction/directive concerning
the quality of the students’ performance, the exact muscle that has to be activated
(tööta reie siseküljega ‘work with the inside of the thigh’). Line 6, however, instructs
the students to terminate the exercise, thus implicitly also launching a transition
to a next one. (The repetition of ja kõverda ‘and flex’ in line 7 is uttered due to
lack of compliance so far.) Likewise, in line 15 the ja-prefaced imperative clause
is uttered to terminate the exercise with the other leg. In contrast to the repetitive
synchronization formulae that are designed to merge with the ongoing movement,
the imperative structures also seek immediate compliance. Instead of accomplish-
ing a continuation of the exercise they launch a change, and that change can also
be partially anticipated based on the customary number of repetitions of this
particular exercise. The instructor can still make mistakes in the count, which
is occasionally commented on in the class, so ultimately the switch to the next
exercise, and especially its timing, is locally negotiated.
Considering the prosodic production of our focus units it seems appropriate
to talk about ja-prefaced units rather than ja as a coordinating device between
clauses. In most instances ja initiates an intonation unit. However, it can also
be prosodically latched to a prior segment of speech, continuing the intonation
unit, as shown in the transcript line 15: the very last formula word is immediately
followed by the conjunction and the transition instruction with the imperative
clause. Together they build a single coherent intonation contour that accompanies
the last move of the exercise and then it immediately goes towards its completion.
This pattern corresponds to the use of ja as a coordinating conjunction before the
last element of a list across spoken and written varieties of Estonian.
The imperative in line 16 directs the students to stay in the position they have
just arrived at and it is uttered without the preface, while again, the instruction
for a next move to lower their hips (ja puusad maha ‘and hips down’) is produced
with the conjunction ja (line 17). In this short excerpt we can thus begin to see the
relevance of ja as a device for temporal extension of a speech formula to achieve
synchrony with the students’ body movements, but also as a device for launching
just-to-be-implemented next moves, in particular together with clausal structures.
While a variety of single words and combinations of a few words can be deployed
in synchronization formulae, which include e.g. imperatives, the immediate next
moves are rather instructed with more elaborate structures, including phrases
(such as puusad maha ‘hips down’) and full clauses. Thus, the linguistic devices of
instruction can be placed on a continuum of more and less elaborate grammatical
structures (see Mondada (2018) on the grammatical structures for instruction in
French), the shortest of which are more useful in formulae and the longest for
transitions between the exercises, while there is of course a gray area between the
two. In general, instruction tends to be maximally indexical, where possible, and
thus concise. Corrections, on the other hand, where there has been a misunder-
standing between the instructor and the students, tend to be elaborate, as can be
seen in the full clausal structures in lines 5 and 9.
In summary, the general distribution of clausal structures in a Pilates class
is that they are deployed at moments of transition between the exercises and in
corrections, while the shorter language structures are more useful for providing
rhythms during the repetitive exercising itself. In particular, Excerpt (2) showed
how the ja-prefaced clausal structures are deployed to arrive at a completion of a
series of exercises (lines 6, 7, 15) by bending the leg back to a relatively less strenu-
ous position. This was also discussed in Excerpt (1), line 5, and can be considered a
regular pattern in the data: ja-prefaced clauses are deployed for a return to various
types of starting positions. It can be sitting, lying down, standing, or the like, and
the common denominator is that they are relatively non-strenuous as opposed to
central parts of the exercise. A ja is thus a first indication that the difficult part
will be over, as we also saw in the immediate reactions by the students above
(Figures 2, 3). At the same time, it provides the opportunity to hold out for a little
more, because there is still time for the actual instruction to emerge. One student
takes this opportunity in the above excerpt. It is also possible for the teacher to
lengthen the ja in cases where the students are persisting in the strenuous posi-
tion. In addition, we saw how a ja-prefaced clausal structure was used to achieve a
transition to a next series of the identical exercise with a different leg (line 8). This,
too, constitutes a regular pragmatic pattern of ja-prefaced clauses.
232 Leelo Keevallik
5. Ja-prefacing and transition to the next sequence of similar exercise moves
As was already pointed out above, one of the main characteristics of the data is
that the production of next clauses without a connector seems to be the default,
since only about one quarter of the clauses were prefaced with a ja (105 as opposed
to 383 without connectors). In the current section we will therefore continue to
explore the specificity of these instances in relation to the default. In addition to
coordinating the termination of an exercise, a ja-prefaced clause seems to provide
the instructions for a next sequence of the same exercise. Excerpt (3) shows a
segment of the class where the students are just about to begin a new exercise.
In line 1 the instructor asks them to turn over onto their stomachs, and in line 2
she provides the label for the upcoming exercise kahe jala löögid ‘two leg kicks’.
The more detailed step-by-step instructions for arms, face, and breathing follow
in lines 4, 5, and 9. None of these three clauses are prefaced by ja; they are instead
produced without verbal connectors. These instructions are preparatory for the
new exercise yet to be performed. Lines 7–8 are more of a meta-comment on the
familiarity of the exercise and thus not part of the instruction.
(3) Cheek to the mat
01 INS: keera ennast siit kõhuli?
turn yourself here onto (your) stomach?
02 (0.3)
03 lähme KAhe jala löögid.
let’s go two leg kicks.
04 paned KÄed siia .h ee selja keskosa peale?
you put (your) hands here in the middle
of (your) spine?
05 -> keerate Ühe põse põra- põrandale matile?
you turn one cheek to the mat on the floor?
06 (1.1)
07 teate MÄletate seda.
(do you) remember this.
08 me oleme seda PAar korda teinud.
we have done it a couple of times.
09 hin:gate siin siss:e?
you breathe in here?
10 (0.5)
11 j:a VÄLja ingates lööd KOlm korda2
and while breathing out you kick three times
12 kandasid istmikule: e vastu. (.)
heels to (your) butt.
2. The first ja-prefaced clause is indeed uttered in line 11, which is yet another pattern discussed
in the following subsection of the paper.
13 nagu ÜK:S:,
like one,
14 (0.4)
15 Kak:s,
two,
16 aga sa ei pea pihta jõudma,
but you don’t have to touch,
17 Kolm,
three,
18 inga sisse siruta KÄed jalad välja
breathe in stretch out (your) arms and legs
19 tõsta rindkere ülesse.
lift (your) upper body.
20 ei käed jäävad külgedele, käed külgedele,
no, arms stay on (your) sides,
arms to (your) sides,
21 nii nagu Reet teeb,
like Reet is doing,
22 ja tee jalad lahti.
and open (your) legs.
23 (0.5)
24 JA::: ALl:a.
and down.
25 -> ja pane teine põsk vastu matti.
and put (your) other cheek on the mat.
26 ja s löö Kokku jälle.
and then kick together again.
27 .h Ük:s,
one,
28 jalg koos.
legs together.
29 Kak:s,
two
30 (0.2)
31 JA: Kolm. (.)
and three.
32 ja:: s:iru:t:a?
and stretch?
33 vaata matile Alla?
look down at the mat?
34 vaata alla vaata alla,
look down look down,
35 -> .h ja teine põsk.
and the other cheek.
36 proovige nii et te ei VEhi jalgadega.
try not to wave (your) legs.
234 Leelo Keevallik
In contrast, when the students are supposed to perform the same exercise on
their other side, the instruction during this transition is prefaced by ja in line 25.
Likewise, the renewed instruction reflects the original one through the choice of
lexicon (one cheek – the other cheek) and the repetition of the necessary location
of their cheek (the mat). It is thus clearly a redoing of the initial instruction in line
5, also linguistically. In line 26 the adverb jälle ‘again’ underlines the fact that the
exercise will be a repetition of the prior. The instructor then launches a count of
numbers to synchronize the practice itself (lines 27–31). In short, the ja-preface
seems to underline the fact that the upcoming task is a next series of the same
exercise, which relates to the discovery by Heritage and Sorjonen (1994) that the
English and-preface is used for formatting a doctor’s question as the next item
on the institutional agenda. Importantly, the instruction to once more change the
cheek in line 35 is again prefaced by a ja. Characteristically, the relatively elaborate
instruction on the first occasion ‘you turn one cheek to the mat on the floor’ (line
5) has now been reduced into an indexical ‘and the other cheek’ (line 35), which
is furthermore sufficient to launch a new series of moves at this point in the class.
Indeed, reduction of instruction on subsequent occasions has been documented in
driving classes in, e.g., Swedish and German (Broth et al. 2017, Depperman 2018).
The excerpt involves other types of ja-prefaced clauses used as part of the
formula for synchronization (line 32) and returning to the starting position (lines
22–24). Many other clauses are corrective comments (lines 16, 20, 33–34, 36) or
further instructions (lines 18–19, 28), using other strategies of coordination. What
was illustrated here is the systematic use of the ja-preface to launch the next iden-
tical series of moves in the same exercise, often with gradually reduced instruction
formats, based on the circumstances that the students are just terminating the
current series. Since during the above excerpt the students are kicking their calves
upwards, the only detail to be changed is their side (cheek), which is also explicitly
mentioned in the reduced versions of the instruction in lines 25 and 35.
As a consequence of the current pattern of launching a next repetition and
since the exits from the exercise can also be indicated with ja-prefaced clauses, the
ja-prefaced structures can cluster around transitions between repetitive exercise
series. This can be observed in lines 22–25: ja tee jalad lahti ja alla ja pane teine
põsk vastu matti ‘and final open (your) legs and down and put (your) other cheek
on the mat’, which guides and accompanies the students through the transition to
a next series of the same exercise in contrast with a repetitive formula that tells
them to go on with whatever they were doing thus far. A ja-preface can thus be a
powerful device for organizing larger activities in a Pilates class, featuring the par-
allelism characteristic of coordinating conjunctions, albeit mostly described in the
syntactic domain (Koivisto 2011). In the current data we see that a ja-preface can
also take care of parallelisms in embodied activities, making audible transitions

between qualitatively similar series of exercises.
6. Ja as a device of cross-modal synchronization and a starting signal
The vowel in the open syllable ja can easily be lengthened and carry a variety of
more or less extreme intonational contours. Accordingly, one of the affordances of
the coordinating ja is that it can delay the production of the upcoming clause and
can thus be a useful device while waiting for students to arrive at the exact position
necessary for the next move. Let us consider Excerpt (4), which illustrates this
capacity of ja in the middle of a complex grammatical structure. The intonation
contour on ja in line 7 iconically reflects the downward body roll.
At the very beginning of the excerpt the initiating moves for the next exercise
are instructed without any verbal connectors. There are altogether four clauses
in lines 1–3 (the last two are produced within a single intonation contour) and
none of them are ja-prefaced. These are consecutive directives for moves to be
accomplished immediately. The students comply with the last instruction during
line 4, after which the instructor starts speaking again, instructing them on their
breathing (line 5) and on their next move in the exercise: ‘open legs as wide as
shoulders’ (line 6). These are yet another two consecutive instructions for the first
instantiation of the exercise series, and as described above in previous excerpts,
these are not prefaced by ja. In line 7, however, the clause rulli alla ‘roll down’
guides the students towards the termination of the exercise. Characteristic of
clauses accomplishing returns to the starting position – which in this case is with
their legs high up in the air (Figure 3) – the current clause is ja-prefaced.
(4) Roll down
01 INS: .hh tõsta Üks jalg õhku,
lift one leg into the air,
02 tõsta teine jalg JUUrde õhku.
lift the other leg up into the air too.
03 inga sin sisse välja ingates rulli jalad üle pea, uhh
breathe in here while breathing out roll (your) legs over (your) head,
04 (1.9)
05 inga SIsse? (.)
breathe in?
06 tee jalad pu- õla laiuselt LAHti, (.)
Open (your) legs as wide as (your) shoulders,
07 -> /N j:a: N/ rulli#1 A::::::::Ll:::::a:::#2.
and roll down.
236 Leelo Keevallik
#1 Students in farthest position, #2 Mid-roll for most

nobody has initiated the roll. students.
08 jalad KOKku#3 ülesse ja:: Üle.

legs together up and over.
#3 Everybody has arrived at the starting position.

09 (0.4)
Crucially, the utterance in line 7 is markedly lengthened and covers at least some of
the time it takes the students to roll onto their backs. Each colon in the transcript
marks the duration of 0.1 millisecond, so it is obvious that the word alla ‘down’
covers most time but the lengthened ja-preface contributes to this clause emerging
as profiled in relation to both prior and ensuing utterances. The overall length
also marks this segment of the exercise as being in focus and perhaps being the
longest and most difficult one. Indeed, it takes time for the students to withhold
the fall onto their backs, as they are arriving only in the middle of line 8, as shown
in Figure 3. In short, the ja uttered with extreme prosody and nasalization does
not only index a return to the starting position of the exercise and its termination
but it also highlights the excessive effort that the students are supposed to exert
at this very moment. Rolling down or up seems to be a frequent way to terminate
a Pilates exercise, and strikingly enough, more than one fifth of our collection of
ja-prefaced clauses expressed this kind of rolling (20 instances out of 105). This,
yet again, reflects the capacity of ja to coordinate consecutive steps of ensuing
physical movement, and furthermore, to achieve a precise match between the
students’ body movements and the online grammar. The ja-prefaced utterances
prompt ensuing moves, one after the other, and we can trace students’ compliance
to each consecutive directive. In this process, ja is a useful device for synchroniz-
ing students so that the instruction emerges precisely before the move itself is
due, or even in partial overlap with it. This particular case also illustrates how it is
possible to match the instructor’s voice quality with the students’ synchronously
expected bodily effort. The bodies of the students and the voice of the instructor
thus come together in this ja-prefaced structure.
Finally, Excerpt (5) shows a similar extendable use of the conjunction between
two clauses but this time the ja functions as a distinct starting signal for the ex-
ercise. The excerpt begins with step-by-step clausal instructions in line 1, guiding
the students to a new exercise, characteristically accumulated without verbal con-
nectors. Interestingly, the conjunction ja is used between two consecutive clauses
in line 2, to launch the exercise after taking an appropriate start position.
(5) In ceiling
01 INS: i:nga sisse, tõsta PArem jalg üles*se. +#1(.)
breathe in, lift (your) right leg up.
ins *R arm arrives
studs +4 legs
lifting off the floor
#1 Instructor showing a leg lift with her arm, four legs lifting off the
floor.
02 INS: J+A::#2.
and.
studs +first 2 legs in place upright
238 Leelo Keevallik
#2 Some legs are in place to start the exercise.
03 INS: *välja ingates lase +PÕlveni? (.)#3

while breathing out let (it) fall
to (your) knee?
ins *R arm falling to chest height
studs *2 more legs arrive in place
+first student: leg at knee height
#3 Instructor showing new leg position with her arm,

three legs at knee height.
04 INS: sisse l*agi, (.)#4

in ceiling,
ins *r.arm upwards
#4 Instructor showing leg up, three students on time, others starting upward.
Lines (1–3) constitute a fully grammatical complex sentence in every sense tõsta
parem jalg ülesse ja välja ingates lase põlveni ‘lift (your) right leg up and while
breathing out let (it) fall to (your) knee’. Breathing cannot be detected in the video
but their legs can, and the transcript reflects the earliest compliances by the stu-
dents to the different instructive segments (see also Figures #1–3). The first ones
start lifting their legs immediately after the instructor’s multimodal instruction,
which involves a verbal and an iconic gestural component – she demonstrates the
leg movement with her arm (as shown in figures #1–4). The short pause in line (1)
and the instructor’s JA:: ‘and’ in line (2) enable the first two students to arrive at the
starting position for the exercise (see Figure #2). We can thus see how the instruc-
tor’s instruction is sensitive to the fact that it takes time for the students’ legs to
arrive in place. The ensuing clause in line (3) is only possible due to the emerging
responsiveness in the students’ bodies, the fact that at least the early performers
have accomplished the move so far (we can see three legs being on time in Figure
#3). The instructor’s verbal explanation as well as her iconic gestures are thus fit-
ted to the trajectories of the students’ bodies, and end up partially overlapping
with them, such as at the end of line (3), where the first student lowers her leg
to knee height in synchrony with the instructor’s PÕlveni ‘to (your) knee’. The
verbal instruction emerges partially as an accompaniment to the body movement,
while it can still function as a prospective prompt for those who are perhaps less
accustomed to the exercise.
Furthermore, besides being a coordinating conjunction within the sentence,
the loud and lengthened JA:: ‘and’ with a distinct falling intonation contour treats
the action so far as being preliminary to the exercise proper, as can be seen from
the teacher’s arm being lowered right afterwards during the instruction for the
students to lower their legs, which is the strenuous part of this exercise. A sharply
240 Leelo Keevallik
falling lengthened ja is a conventionalized device for marking a start, and thus

an essential device for practice coordination (the item ‘and’ can furthermore be
used in different languages for this very function, see e.g. for Swedish in Broth and
Keevallik 2014, which is also an additional hint that the Estonian ja in this case
is not an affirmative response). It is a major juncture marker in the activity and a
starting signal for the actual exercise. Even this type of ja crucially structures the
activities during the class, albeit in yet a different way. Equally distinctive prosodic
variants of the English and have been shown to achieve different functions, so
that the most salient one is used for the largest scope in conversation (Barth-
Weingarten & Couper-Kuhlen 2011, Barth-Weingarten 2012). The start-signal
usage of the Estonian conjunction ja features an independent intonation contour
that falls in a distinct steep curve from high to low. It is thus not the preface of an
intonation unit, in contrast to the patterns discussed above for returns to starting
positions and transitions between series of parallel exercises.
7. Conclusion
This paper has argued that clausal coordination with ja is used in various ways to
organize and synchronize activities in a Pilates class: to return to a less strenuous
position and terminate the exercise; to launch an identical series of moves in an
exercise with a different leg, on the other side, or the like; and for the temporal
extension of talk together with opportunities for highly expressive and iconic into-
nation. A ja with a separate intonation contour featuring a distinct high-low fall is
furthermore used as a starting signal for the actual exercise. Complex coordinated
sentences uttered by the instructor guide the students through transitory segments
of the class, moving from one exercise to the next, or launching the next exercise
sequence in a series. Grammar was thus shown to emerge across participants and
modalities, in a fine-grained attunement to the recipients’ bodies. The instructor
could, for example, extend a ja while waiting for the students to arrive at a certain
position for her next utterance to be appropriately timed with the next move. We
were thus able to witness moments of designed synchrony between speech and the
exercising bodies, as well as dissect the constituent functions of the coordinating
conjunction ja. The conjunction emerges as a temporal index in the embodied
trajectories of action.
In contrast to the short synchronizing formulae that are used to achieve repeti-
tive rhythms (Keevallik, in press), the use of ja as a coordinating conjunction before
clauses instead happens in transitions between exercises where the instructor’s
objective is to guide the students’ bodies through locally new moves and positions
in order to launch a different exercise. This is also where the predictability of every
next move for the students is lower than when in the middle of a series of repeti-
tions. This results in more local responsiveness by the students to the instructor’s
talk as opposed to just carrying on with the exercise. Overall, the participants can
be observed to jointly and locally construct every next step through a coordina-
tion of talk by one participant and physical moves by others.
Temporal coordination is a constant challenge for the Pilates instructor, as
it is often merely by the use of her voice, grammar, and gesture that she guides
the students through the classes. Only occasionally does she perform the moves
herself. The details of her grammar together with the prosody and voice modu-
lations (such as nasalization) are often the only source of information for the
students, in particular when they are lying on the floor in various positions that
make observing the instructor and her gestures impossible. As was shown above,
the students tune in to the minute details of the instructor’s vocal production.
This is possible partly because the exercises are predictable to them, but the actual
transitions and especially their timing are always a local accomplishment, which
all the participants, including students, may comment on at moments of failure.
The students can be seen to adjust their moves and pace to the instructor’s utter-
ances containing the conjunction ja, not only when it is part of repetitive formulae
that synchronically accompany an exercise, but also when it is used in transitory
segments between the exercises or as a starting signal for an exercise proper. In
this multi-party activity setting coordination of clauses accordingly emerges in
real time responsively to the recipients’ bodily moves. Grammatical coordination
can thus be seen as an epiphenomenon of interaction between the instructor and
the exercising students.
While a lack of clause-initial connectors is the default in this activity, the
explicit conjunction ja is certainly meaningful in notifying the students of an
upcoming relief from a strenuous position, a return to the starting position, or
launching a next sequence in a series of an exercise. Through its prosody it can
wait for the students’ bodies to reach a next due movement and thereby display
an understanding of what the students are experiencing simultaneously in their
exercising bodies. The understanding is essentially embodied, displayed through
vocally exerted energy. All of this is an immense accomplishment for a word that
has occasionally been characterized as semantically “void”.
Acknowledgment
This research was carried out within the project “Vocal Coordination of Human Action”, funded
by the Swedish Research Council 2016–00827.
242 Leelo Keevallik
Transcription conventions
underlining – stress
CAPS – emphasis
[] – overlaps
= – latching of units
(0.5) – pause length in tenths of a second
(.) – micropause
h – breathing out
.h – breathing in
((comment)) – transcriber’s comment
: – lengthening of a sound (each : corresponds to 0.1 sec)
. – pitch fall at the end of an intonation unit
? – pitch rise at the end of an intonation unit
, – level pitch at the end of an intonation unit
<slower> – slower speech
>faster< – faster speech
// – special voice quality
@ – smiley voice
& – tense glottis, displaying strained body
N – nasal voice
Q – various other qualities
° – sotto voce
↑ – distinct pitch movement upwards
*+ – alignment of bodily moves
References
Barth-Weingarten, D., & Couper-Kuhlen, E. (2011). Action, prosody and emergent construc-
tions: The case of ‘and’. In P. Auer, & S. Pfänder (Eds.), Constructions: Emerging and emergent
(pp. 264–293). Berlin: De Gruyter Mouton. https://doi.org/10.1515/9783110229080.263
Barth-Weingarten, D. (2012). Of Ens ‘n’ Ands: Observations on the phonetic make-up of a
coordinator and its uses in Talk-in-Interaction. Language and Speech 55(1), 35–36.
https://doi.org/10.1177/0023830911428868
Broth, M., Cromdal, J., & Levin, L. (2017). Starting out as a driver: Progression in instructed
pedal work. In Å.Mäkitalo, P. Linell, & R. Säljö (Eds.), Memory practices and learning:
Interactional, institutional and sociocultural perspectives (pp. 115–152). Charlotte, N.C.:
Information Age Publishing.
Broth, M., & Keevallik, L. (2014). Getting ready to move as a couple. Accomplishing mobile
formations in a dance class. Space and Culture 17(2), 107–121.
https://doi.org/10.1177/1206331213508483
Deppermann, A. (2018): Changes in turn-design over interactional histories – the case of

instructions in driving school lessons. In A. Deppermann, & J. Streeck (Eds.), Time in em-
bodied interaction: Synchronicity and sequentiality of multimodal resources (pp. 293–324).
EKSS – Eesti keele seletav sõnaraamat [Comprehensive dictionary of Estonian]. http://www.eki.
ee/dict/ekss/ Retrieved 20.05.2018
Erelt, M., Kasik, R., Metslang, H., Rajandi, H., Ross, K., Saari, H., Tael, K., & Vare, S. (1993).
Eesti keele grammatika II (Estonian grammar II). Tallinn: Eesti Teaduste Akadeemia Eesti
Keele Instituut.
Hakulinen, A., Vilkuna, M., Korhonen, R., Koivisto, V., Heinonen, T., & Alho, I. (2004). Iso
suomen kielioppi (The Comprehensive Grammar of Finnish). Helsinki: Suomalaisen Kirjal-
lisuuden Seura.
Haspelmath, M. (2007). Coordination. In T. Shopen (Ed.), Language typology and syntactic
description, vol. II: Complex constructions (pp. 1–51). 2nd ed. Cambridge: Cambridge
Hennoste, T., Lindström, L., Rääbis, A., Toomet, P., & Vellerind, R. (2000). Eesti suulise kõne
korpus ja mõnede allkeelte võrdluse katse [The corpus of spoken Estonian and an at-
tempt at comparing some varieties]. In T. Hennoste (Ed.), Arvutuslingvistikalt inimesele
(pp. 245–283). Tartu: Tartu ülikooli üldkeeleteadus õpetooli toimetised 1.
Heritage, J., & Sorjonen, M.-L. (1994). Constituting and maintaining activities across sequences:
And-prefacing as a feature of question design. Language in Society 23(1), 1–29.
https://doi.org/10.1017/S0047404500017656
Kalliokoski, J. (1989). Ja: Rinnastus ja rinnastuskonjuktion käyttö [Ja. Coordination and the use
of the coordinating conjunction in Finnish.]. Helsinki: Suomalaisen Kirjallisuuden Seura.
Keevallik, L. (2000). Keelendid et ja nii et vestluses. [Linguistic units et ‘that’ and nii et ‘so’ in
conversation.] Keel ja Kirjandus 43(5), 344–358.
Keevallik, L. (in press). Synchronizing instruction and compliance: The emergence of gram-
matical structure in a Pilates class. In C. Taleghani-Nikazm, E. Betz, & P. Golato (Eds.),
Mobilizing others Grammar and lexis within larger activities. Amsterdam, Philadelphia:
John Benjamins.
Koivisto, A. (2011). Sanomattakin selvää? Ja,mutta ja että puheenvuoron lopussa [Goes without
saying? Finnish conjunctions ja, mutta and että in turn-final position]. Helsinki: Helsingin
yliopiston suomen kielen, suomalais-ugrilaisten ja pohjoismaisten kielten ja kirjallisuuks-
ien laitos.
Local, J. (2004). Getting back to prior talk: And-uh(m) as a back-connecting device in British
and American English. In E. Couper-Kuhlen, & C. Ford (Eds.), Sound patterns in interac-
tion (pp. 377–400). Amsterdam, The Netherlands: John Benjamins.
https://doi.org/10.1075/tsl.62.18loc
Mazeland, H., & Huiskes, M. (2001). Dutch ‘but’ as a sequential conjunction: Its use as a resump-
tion marker. In M. Selting, & E. Couper-Kuhlen (Eds.), Studies in interactional linguistics
(pp. 141–169). Amsterdam, The Netherlands: John Benjamins.
https://doi.org/10.1075/sidag.10.08maz
Mondada, L. (2018) Driving instruction at high speed on a race circuit: Issues in action forma-
tion and sequence organization. International Journal of Applied Linguistics 28, 1–22.
Nevile, M. (2006). Making sequentiality salient: And-prefacing in the talk of airline pilots.
Discourse Studies 8(2), 309–332. https://doi.org/10.1177/1461445606061797
244 Leelo Keevallik
Payne, T. E. (1997). Describing morphosyntax: A guide for field linguistics. Cambridge: Cam-
bridge University Press.
Schiffrin, D. (1987). Discourse markers. Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511611841
Turk, M. J. (2004). Using and in conversational interaction. Research on Language and Social
Interaction 37(2), 219–250. https://doi.org/10.1207/s15327973rlsi3702_5
Chapter 9
Consecutive clause combinations in

instructing activities
Directives and accounts in the context of
physical training
Jan Lindström, Camilla Lindholm, Inga-Lill Grahn and

Martina Huhtamäki
This chapter investigates the formatting of instructions in physical training

with personal trainers or physiotherapists. Instructions occur in multimodal
activities where invitations to action, compliances with them, and accounts
for them emerge through grammatical, prosodic and embodied resources. We
identified a two-part pattern [directive & account] that accomplishes a complex
structural and pragmatic unit in trainers’ instructions. The instructions are
grammatically formed of consecutive clause combinations in which the directive
part is a declarative or an imperative. These combinations emerge in interactive
sequences and are a designed, rather than a contingent feature in the making of
instructions. Nevertheless, there is variation in their sequential emergence and
grammatical and prosodic composition, from tight packages to projected or
expanded clause/action combinations.
Keywords: directives, instructions, accounts, clause combining, action

combination, emergent grammar, projection, expansion, prosody, embodiment
1. Introduction1
Instructing another person and learning from one another constitute central parts
of human interaction and culture. Instructing activities rely on a combination of
1. The present study was supported by a grant from Riksbankens Jubileumsfond for the re-
search program Interaction and Variation in Pluricentric Languages – Communicative patterns in
Sweden Swedish and Finland Swedish (grant nr. M12-0137:1) and by a grant from the Academy
246 Jan Lindström, Camilla Lindholm, Inga-Lill Grahn and Martina Huhtamäki
verbal language and visual, also physical, demonstrations through which skills are
passed on to a new generation in different forms of expert–novice interaction.
Needless to say, instructing is not a business of mere commanding and issuing
simple imperatives. Pedagogical studies show that students often lack interpreta-
tive resources for seeing the purpose of a task (see, e.g., Lindwall, 2008; Lindwall &
Lymer, 2014); hence, students are probably more likely to focus on a learning task
when they understand the goal of the instruction (see, e.g., Fisher & Frey, 2014).
Unmotivated instructions can, at worst, lead to breakdowns of intersubjectiv-
ity between the instructor and instructed (see Deppermann, 2015). Instructions,
then, can occasion problems of clarity, consistency, adequacy, completeness, and
followability (cf. Garfinkel, 2002: p. 198).
In this study, we investigate instructions in Swedish talk-in-interaction that
take the form of a combination of two clauses and actions expressed with them.
The analysis concentrates on interactional sequences involving two participants:
physiotherapists or personal trainers and their clients doing or preparing to do a
physical exercise. The trainer2 instructs the client about the exercise and the effects
of a certain training program or workout. Such instructing activities involve direc-
tives, explanations, motivations and conclusions addressed to the client and these
are often expressed in multi-clausal turns. Here we set out to analyze one such
formal pattern that repeated itself in our data: it consists of a directive clause (for
example, in the imperative) that is followed by an adverbial, consecutive clause
that accounts for the purpose or result of the nominated action. It is this two-part
pattern, [directive & account], that accomplishes a clause combination as well as an
action combination (see Couper-Kuhlen & Etelämäki, 2014, 2017 for the concept).
Such action combinations are labeled as instructions in our analysis. These instruc-
tions emerge via a coordination of linguistic and multimodal practices among the
participants throughout the instructional sequence and evidence how grammar
is fitted to sequences and trajectories of embodied activities (Keevallik, 2018; see
also Reed & Szczepek Reed, 2013 for instructions as multi-layered interactional
projects).
of Finland for the project Emergent Clausal Syntax for Conversation: Swedish in a cross-language
comparison (grant nr. 316865). We want to thank Leelo Keevallik and Simona Pekarek Doehler
for their valuable comments on an earlier version of this text. We also thank our research as-
sistants Madeleine Forsén and Susanna Siljander for help with data excerpting.
2. We use the label ’trainer’ in the text (and T in the excerpts) as a generic reference to the
instructor, both in case of physiotherapists and personal trainers. ‘Client’ (and C in the excerpts)
refers to the trainees.
Chapter 9. Consecutive clause combinations in instructing activities 247
2. Instruction as an interactionally emerging phenomenon
In order to see how instructing sequences emerge in practice, we can turn to

Excerpt (1), where the trainer (T) guides the client (C) verbally and physically. The
trainer assists the client to stretch her leg muscles by holding the left leg in an up-
ward position while the client is lying on her back on the floor, with her arms along
the body side and the hands palm down. To keep the body relaxed, the trainer
asks the client to turn the palms to face upward (sätt ännu handflatan ner eller
liksom upp mot taket ‘turn your palm down, I mean, upward towards the ceiling’,
l. 5–6). The client complies and the trainer acknowledges this with a “third” (see
Keevallik, 2018), upon which she produces a consecutive purpose clause så att den
där axeln får ‘so that the shoulder gets [relaxed]’ that accounts for and completes
the instruction (l. 10); see Figure 1 for the illustrations of the embodied acts. (Note
that the lines in focus are bolded and highlighted with grey shading both in the
original and translation: bold=directive, bold & grey=account.)
Figure 1. Client lying down, palms down (left); trainer demonstrating palm up (center);
client lying down, palms up (right).
(1) ’A stretch’ (PTR:004II:43).

01 T: Å *så tar ja lite å tänjer.
And then I will go on and stretch a little
*T lifts C’s leg straight upwards and stretches
02 (0.6)
03 T: Å nu försök slappna av.
And now, try to relax
04 (.)
05 T: Sätt ännu eh handflatan ne:,
put.imp yet prt palm.def down
Turn your um palm to face downward as well,
06 eller liksom *upp mot *take.
or prt up toward ceiling.def
I mean, upward towards the ceiling.
*T turns her palm up
*C turns her palm up
07 (0.5)
08 T: Jep.
Yup.
09 (0.2)
10 T: Så att den där axeln °får°
so that that.dem shoulder.def get.prs
So that the shoulder gets [relaxed]
When issuing the directive in lines 5–6, the trainer simultaneously demonstrates
the requested palm position with her own hand, thus illustrating the verbal direc-
tive (and repairing her initial wording “to face downward”, l. 5). The client’s physi-
cal compliance, turning the palm upwards, is acknowledged by the trainer (l. 8)
and as such is a sequence-closing move (see Keevallik, 2018: p. 3–5). Nonetheless,
the trainer expands the sequence by adding an account with the consecutive clause
in line 10, which, in effect, redresses the plain directive in line 5–6 as an instruc-
tion (with the components [directive] & [account]). The preceding directive was
delivered with final falling intonation. However, the account-giving expansion
is linked to the directive as an increment (see Couper-Kuhlen & Ono, 2007 for
increments): it is fitted in with the subordinating conjunction så att ‘so that’ and
prosodically delivered without pitch reset in the onset. This subsequent unit is also
pragmatically projected because the initial directive, which concerns hand posi-
tion, leaves the question ‘why that now’ open, as the trainer is presently manipulat-
ing the client’s leg. The temporal gap between the directive and the account can, in
the end, result from the trainer seeking to coordinate talk with bodily movement
(cf. Goodwin, 2000; Mondada, 2009), as the verbal and embodied actions are car-
ried out in response to each other: the client’s successful compliance, so to speak,
enables the trainer to complete the sequence with an account.
From a post hoc semantic-grammatical point of view the instructional se-
quence in (1) contains a consecutive clause combination, as represented in Table 1,
although the combination in practice emerges across a sequence of actions: direc-
tive–compliance–acknowledgement–consecutive.
Table 1. Instruction in Excerpt (1)

Excerpt Clause 1 Combining Clause 2
Directive element Account
(Consequence)
(1) Sätt ännu eh handflatan ne:, eller Så att den där axeln
liksom upp mot take. °får°
Turn your um palm to face downward, I So that the shoulder gets
mean, upward towards the ceiling. [relaxed]
As regards their sequential composition, combinations of directives and accounts

can materialize through the processes of expansion (as above), projection, or single-
packaging in one intonational-structural unit.
Our study focuses on the issues mentioned above regarding the temporal,
structural and interactional emergence of instructions in the context of physical
training. By doing this, we add to the plethora of research on requesting in social
interaction (see, e.g., Drew & Couper-Kuhlen, 2014), in the sense of ‘requesting
action’. We call the actions and activity taking place in our cases ‘instruction’ and
‘instructing’ rather than request or requesting in order to stress the pedagogical
aspects involved: the instructions are produced in a context where the other party is
expected to learn how to carry out certain forms of exercise, and not, for example, to
assist the requester. Also advice-giving is a broader phenomenon than instructing
and may incorporate such less binding actions as counseling, recommending and
information-giving (see Limberg & Locher, 2012). As an investigation of instructing
and directing, our study contributes to research on directives (see, e.g., Ervin-Tripp,
1976; Sorjonen, Raevaara & Couper-Kuhlen, 2017) and instructional interaction
(Szczepek Reed et al., 2013). But most specifically, we follow an interactional linguis-
tic agenda by investigating how linguistic structure can be related to interactional
function (see Couper-Kuhlen & Selting, 2018), placing combinations of clauses and
actions at center stage (see Couper-Kuhlen and Etelämäki, 2014, 2017).
In this vein of interactional studies of instructing activities, Keevallik (2013,
2015) points to cases where instructions consist of a projecting clause followed by a
physical teacher demonstration that completes the projection and the instruction.
The temporality of the instruction and the expected response is central also in our
study, in the sense that compliance with instructions is expected to be produced
here and now. We can also see a parallel in the work on driving lessons by De
Stefani and Gazin (2014), who distinguish between instructions that have to be
immediately complied with, and instructions with a longer temporal scope (see
also Reed & Szczepek, 2013). We follow up the dialogical trajectories in verbal and
embodied instructing sequences, focusing on local actions in the instructional
activity. Instruction can also be analyzed in a more holistic way so that instruct-
ing is viewed as a project with characteristic practices inhabiting the beginning,
middle and end (see Reed & Szczepek, 2013); such a broader investigation of what
instructing is built of does not fit in the limitations of our structural focus.
As was shown in Excerpt (1), instructing is an interactionally emerging pro-
cess in which an understanding of what to do and for what purpose is established
between two participants. The process often involves a demonstration of a physi-
cal movement and the client’s repetition of the movement that complies with the
directive to act. In other cases the client is performing an exercise and the trainer
intervenes by giving a directive to alter or correct the physical performance.
The study is organized as follows. To begin with, we will briefly describe our
data in Section 3 and the collection of instances that this study is based on. Section 4
provides an overview of the structural patterns of instructing clause combinations
in the data. Section 5 then provides detailed sequential analyses that account for
how the structural emergence of instructing is intertwined with the intersubjective
acts of directing and complying. Since the activity context revolves around physical
actions, the analysis of structural emergence will keep track of multimodal aspects
of the participants’ orientation to the formulation of instructions. The chapter
ends with a summary that details grammatical and prosodic features of the clause/
action combinations discussed in the analysis, followed by a concluding section.
3. Data and collection
Our data consist of 14 hours of video recordings of training sessions with per-
sonal trainers/physiotherapists and their clients. The participants are engaged in
physiotherapeutic treatment or preventive healthcare activities, that is, the clients
have turned to the experts in order to get and stay fit and healthy. During a session
the trainer gives full attention to the client by instructing and encouraging him/
her according to a personalized treatment or training program. Personal training
also includes access to health clubs dedicated to physical activity with a variety of
training equipment both for training and for relaxing after sessions. Recordings
were made in Finland (12 sessions, 10 hours of data) and Sweden (3 sessions, 4
hours of data) and were transcribed according to general CA conventions (see,
e.g., Ochs, Schegloff & Thompson, 1996). Embodied actions that are referred to or
consequential for the analysis are notated in the transcripts.
The two sub-corpora are similar with regard to the language used (Swedish),
institutional roles (personal trainers/physiotherapists and clients), activity type
(training, instructing) and the period of data collection (between 2016 and 2017).
The primary focus in the training is on physical performance and embodied
compliance and not on a verbal display of learning (see also Reed & Szczepek
Reed, 2013 on music masterclasses). Out of this data, instructing sequences were
collected according to the following criteria: 1) the trainer expresses a directive to
act, 2) the directive is elaborated with a consecutive clause expressing purpose or
result, and 3) the client displays verbal or non-verbal compliance or understanding
(after the directive and/or the elaboration). The collection consists of 22 instances
in the data from Finland and 21 from Sweden.3
3. Despite the fact that the sub-corpora from Finland and Sweden respectively are different in size,
they happened to yield an almost equal number of instances according to the collecting criteria.
The data were collected as a part of a larger research program on Swedish as

a pluricentric language, Interaction and variation in pluricentric languages (IVIP)
(see Norrby et al., 2012). The program conducts empirical studies of interaction in
Swedish in Sweden and Finland and uses interactional linguistic methods, com-
munication ethnography and a variational pragmatic framework to examine the
complex relationship between language, nation, and culture. As the data represent
two distinct, national varieties of Swedish, which reportedly differ in some as-
pects of pronunciation, prosody, lexis, syntactic construction and pragmatics (see
Norrby et al., 2012; Reuter, 1992; Wide & Lyngfelt, 2009), we will also address the
question of linguistic variation when it becomes relevant in light of our present
collection of instances.
4. Overview of the instruction formats
The instructing actions we have collected in the data are composed of two clausal
components: an initial directive component and a subsequent account-giving
component that stands in a consecutive relation to the initial component. Together
they form a clause combination in which a consecutive conjunction or an adver-
bial connective does the lexical combinatory work. The two parts communicate
separate actions, namely directing and accounting respectively, but in combina-
tion with each other they accomplish one complex action, an instruction. The
clause combination is not of the classical type, such as between a [matrix clause
& complement clause] or an [antecedent & consequent] in conditional clause
combinations, as a strictly structural dependency relation does not exist, for ex-
ample, between an imperative clause and a consecutive clause. Nonetheless, it can
be argued that there is a functional dependency relation between the component
clauses: the directive is always the grammatical main clause, whereas the consecu-
tive is constructed as a subordinate clause or as a [main] clause that is structurally
recognizable as a “subsequent”, i.e. bearing marks of not standing alone (see J.
Lindström, 2014a; also Günthner, this volume).
The directive component is expressed with a declarative or imperative clause
or sometimes also with formally subordinate clauses or clausal fragments, e.g., å
tårna ba lätt imot ‘and the toes only lightly against’, cf. ‘and [keep] the toes only
lightly against [the wall]’ (Grahn & Huhtamäki, 2019). As shown in Table 2, the
typical directive is a main clause, though, and declaratives are slightly more com-
mon than imperatives.
Table 2. Distribution of grammatical formats in the directive part of instructions

Directive form Occurrences
Declarative 20
Imperative 15
Other 8
Total 43
Declaratives are usable as directives because they can be modulated in various

ways; for example, by including the second person pronoun to highlight the other
(client) as agent (2a), by including the first person pronoun to highlight self (train-
er) as the agent (2b), or by using an existential format to mitigate the agent (2c).
(2)4 a. Du kan ta ännu bollen närmare.
‘You can take the ball even nearer.’
b. Jag vill att du har lika mycket kraft nu i varje ben.
‘I want you to have as much power now in each leg.’
c. Det är också lättare att hålla andra handen lös.
‘It is also easier to keep the other hand loose.’
As in Excerpt (1), directives in the imperative are expressed directly, without

any modulation with polite formulae or mitigating or intensifying particles (see
Lindström et al., 2017 for imperatives and their modulations in Swedish medical
consultations). Nonetheless, there is one special feature in the distribution of imper-
ative formats: six out of seven imperatives in the data from Sweden are constructed
with the formula tänk på (att …) ‘consider (that …)’, whereas this type of imperative
does not occur in the data from Finland. Excerpt (3) shows this directive format:
the trainer and client are standing, facing each other, and the trainer intervenes in
an exercise the client has been doing with a ball. This instruction, then, introduces a
correction: the trainer points out how the ball should be handled, also demonstrat-
ing this with swaying arm and body movements (without a ball) from side to side.
(3) ’The ball’ (PTR:02:2).
01 T: N[u, (0.3) har du] bollen *här ganska nä[ra dig?]
Now, you have the ball here rather close to you
*T sways arms to her sides
02 C: [A jo ]
Oh yes
03 C: [Ja jo ]
Yes right
4. These examples are culled from our collection but are somewhat normalized as regards spell-
ing and sequential flow for a better overview of the basic grammatical structure.
04 (0.3)
05 T: *Tänk på att du kan ha bollen
think.imp prt that you can have.inf ball.def
Consider that you can have the ball
*T makes big movements with her arms from side to side
06 ännu mer, (.) [så att] du får=
yet more so that you get.prs
even more (to your side), so that you’ll get
07 C: [jo ]
yes
08 =ännu mer rotation.
yet more rotation
even more rotation.
The imperative with tänk på (l. 5) not only invites action but also introspection,
and it therefore places the client closer to center stage as an active co-participant
who is able to monitor and evaluate the implementation of the exercise. We can
see here a link to a more pronounced person orientation in institutional discourse
in Sweden; for example, Norrby, Wide, Lindström and Nilsson (2015) found that
T-address was abundant in Sweden-Swedish medical consultations, whereas im-
personal formats were more characteristic of Finland Swedish, and Henricson and
Nelson (2017) found that Sweden-Swedish academic supervision meetings were
more concerned with participant relations than corresponding Finland-Swedish
meetings that focused on the task at hand. The tänk på-format also involves some
softening of the directive, which in (3) redresses a corrective move as a mere tip
that could be considered. Instructions in the tänk på format are somewhat more
complex from the point of clause combination, because the introspection-inviting
introductory part of the directive has the role of a matrix sentence that takes the
action invitation as a complement clause: [tänk på] & [att du kan ha bollen ännu
mer]. This bi-clausal directive, then, is further combined with a så att clause that
is a subordinate purpose clause: {directive [tänk på] & [att du kan ha bollen ännu
mer]} & {account [så att du får ännu mer rotation]}.
In our data, the consecutive, account-giving component is combined with
the directive via the subordinating conjunctions så and så att ‘so’ or with the
adverbial connective så ‘so, and’ that is followed by a main clause with an inverted
word order5 (finite verb + subject). Hence, the action combination [directive &
account] can be realized through hypotactic or paratactic clause combination (see
5. When så is an adverb, and not a subordinator, it occupies the “front field” position of a clause
(see, e.g., J. Lindström, 2014b). Because Swedish is a Germanic V2 language, the adverbial front-
field constituent must be followed by the finite verb, and then by the subject: så ser du det ‘so you
see it’, lit. ‘so see you it’.
Matthiessen & Thompson, 1988). There is a slight semantic difference between

these in that the subordinate clauses convey the purpose of an action (what can be
achieved through it; Excerpts 1 and 3 above), while the så-clauses with main clause
syntax focus on the result (what the effects of an action are; Excerpt 5 below).
As can be seen in Table 3, purpose clauses are more frequent in the instructions.
Even negative purpose or result is expressed in some instances (så att du inte bara
stannar ‘so that you don’t just stand in one place’), that is, the client is directed to
act in a certain way to avoid an unfavorable consequence (see Excerpt 4 below).
Furthermore, the consequence can be agentive, focusing on what the client can
achieve (så att du får ännu mer rotation ‘so that you’ll get even more rotation’), or
it can be expressed in a generic way (så att de blir lite mer flow i de ‘so that there will
be more flow in it’), defocusing the active subject (see Limberg & Locher, 2012).
Table 3. Distribution of connectors introducing the consecutive clause

Connector Occurrences
så (adverb); result 12
så (conjunction); purpose 11
så att (conjunction); purpose 20
Total 43
Instructions occur in different phases of the physical exercise in focus, considering

whether they are issued before or during the client’s performance of the exercise.
Table 4 gives an overview of this phasal patterning of instructions in the data: 1)
preparatory instructions in which the trainer says and demonstrates him-/herself
what the client is supposed to do; 2) preparatory instructions in which the physi-
cal demonstration involves the client’s body as a point of reference, for example,
through touching a focused body part, and 3) corrective instructions that respond
to how the client is performing an exercise. The counts refer to occurrences of
instructions in each phase.
Table 4. Distribution of instructions related to the phase of exercise

Type of instruction Occurrences
Preparatory; T self-demonstrates 16
Preparatory; T involves C in demonstration 11
Corrective; T intervenes in C’s performance 16
Total 43
In the following, we will proceed with a closer analysis of some representative

instances of instruction types according to their occurrence in the training phases.
5. Analysis of instruction sequences
This section offers analyses of instructions as emergent sequential, interactional

processes in which the practices of clause combining, the participants’ initiatives
and responses, and their embodied acts are intertwined. The organization of the
analytic sections is based on where the type of training instructing occurs in the
activity, i.e. as preparatory or corrective actions (see Table 4 above). The first
two phases include preparatory instructions in which the trainer either makes
a demonstration him-/herself or involves the client in the demonstration. The
third category involves cases in which the instruction occurs while the clients are
performing exercises and is produced to correct the clients’ performance.
5.1 Preparatory instructions: Demonstration by trainer
In this category the instructing sequences contain a physical demonstration from

the trainer as part of an instruction when introducing a new exercise, a variation
of which was seen in Excerpt (3). The consecutive clause and the physical dem-
onstration complete the instruction and account for it. Excerpt (4) contains two
consecutive clauses produced after a directive action that is complemented with a
physical demonstration by the trainer. The consecutives account for what should
be achieved (or not achieved) through implementing the exercise in the instructed
way. The trainer and the client are standing facing each other while the trainer
introduces a squat-and-jump type of exercise.
(4) ’Leg bending’ (ptr:02:15).
01 T: När du gör den med benböjen, (0.7)
when you do-prs that with leg.bend-def
when you do the one with leg bending,
02 så vill ja att (0.3) nu (.)
prt want I that now
I want that (you) now
03 att du *#gör dom här tre där¿
that you do-prs those-dem three there
that you do these three (like this)¿
*T squats
#Figure 2 left
04 (0.5)
05 T: *Så en två tre #upp* (.) ner.
So one two three up, down.
*T squats stepwise and ends with a jump on “upp”*
#Figure 2 right
06 (0.4)
07 T: *En två tre (0.4) °upp°.*

One two three, up.
*T squats stepwise and ends with a jump on “upp”*
08 T: *så att de blir lite mer flow i de.
so that it become-prs a.little more flow in it
so that there will be a little more flow in it
*T gesticulates with her hands up and down
09 C: Ja oke[j ]
Yes, okay
10 T: [Så] att du inte ba *stannar,
so that you not only stall-prs
So that you don’t just stand in one place
*T stands in a frozen position
11 (.) å sen så,=
and then prt
and then,
12 C: =Absolut (.) okej.
Absolutely, okay.
Figure 2. Trainer demonstrates squat down (left); Trainer demonstrates jump up (right).
The exercise is referred to by the trainer as den med benböjen ‘the one with leg
bending’ (l. 1). The client is supposed to squat, bending his legs, three times and
end with a jump in an upright position, and then to repeat this ten times. The
instruction begins with a declarative-formatted directive (l. 1–3) from the trainer
who is getting into position with slightly bent knees: så vill jag att nu (.) att du
gör dom här tre där ‘I want that now (.) that you do these three’. The declarative
is followed by two bodily demonstrations showing how to ‘do these three there’,
both accompanied by a phrasal counting unit en två tre upp ‘one two three up’
(l. 5 and 7; Figure 2). The demonstration explicates the physical execution of the
performance, while a verbal description is not produced in this slot (see Keevallik,
2015 for verbal projection and embodied completion).
After the demonstrations, the trainer resumes a standing position and pro-
duces a consecutive clause, in which she explains that the method demonstrated
gives more “flow”: så att de blir lite mer flow i de ‘so that there will be a little
more flow in it’ (l. 8). The consecutive clause is combined with the directive and
demonstrating part of the instruction with the conjunction så. This account-giving
part contains the formal subject de ‘it’ and the verb bli ‘become’, which gives the
achievement of “flow” an impersonal aspect. The client displays understanding
with ja okej ‘yes okay’ in line 9 and takes a couple of steps towards the gym mat on
which he should perform the exercise. The trainer then continues on with another
consecutive clause that offers an alternative account: it is negated and contains the
subject du ‘you’ (l. 10), thereby moving the focus to the client as an agent in telling
him what he should not do: så att du inte ba stannar ‘so that you don’t just stand in
one place’. The personal orientation prompts the client to start the exercise, while
the client confirms his understanding of the instruction and willingness to comply
with an upgraded receipt in line 12: absolut (.) okej ‘absolutely (.) okay’, then mov-
ing on to do the exercise.
To form a directive concerning another person’s bodily movement by using
the subjective volitional format jag vill att du ‘I want you to’, is not uncommon in
our data. It could be heard as quite self-centered, even rude, in another situation,
but in this specific instructing context it emphasizes the trainer’s deontic authority
as a professional who is responsible for the training (see Stevanovic & Peräkylä,
2012). The format with the volitional verb also projects the upcoming core of the
directive, possibly involving a longer spate of instruction (see Reed & Szczepek
Reed, 2013: p. 321), which comes in the form of physical demonstration and
counting in lines 5–7. The embodied demonstrations and the ensuing consecutive
clauses are mutually meaning-making and carefully timed: the consecutives are
expressed only after the physical demonstration is fully concluded. This sequential
emergence of the clausal and (inter)actional structure between the trainer (T) and
client (C) can be outlined in the following steps:
1. T: Directive inviting action (declarative clause)
2. T: Demonstration of invited action (embodied, verbal support)
3. T: Account for the directive in a consecutive clause
4. C: Verbal display of understanding
5. T: Account for the directive in a negated consecutive clause
6. C: Verbal display of understanding
The steps 1) and 3) in this instruction sequence are chunked prosodically into units
of their own, separated by the demonstration in step 2). Nonetheless, the consecu-
tive component is produced prosodically as a continuation, as there is no final fall
in the directive component (l. 3 in Excerpt 4) and no upstep at the beginning of the
consecutive part (l. 8). This establishes the embodied demonstration as a designed,
integrated part of the directive, and not as an inserted, separating segment.
In summary, the trainer’s instructing turn unfolds in a temporal progression out
of three distinctive parts – directive, demonstration, account – that are sequenced
after each other to organize and coordinate social action. This accomplishes a
single complex action which, in a post hoc structural perspective, is a complex con-
secutive clause combination that in its process of emergence is matched with the
participants’ initiating and responding moves in the local instruction sequence.
The consecutive clause completes this complex action combination pragmatically
and the recipient takes an orientation to it as one of completion.
5.2 Preparatory instructions: Client involved in trainer’s demonstration
The instances in this category differ from the type of instructions discussed in
section 5.1 in that the clients here are involved in the trainer’s demonstrations.
Excerpt (1) presented one such instance: the trainer proceeds with the directive
and consecutive components, and the client does the directed action while being
instructed. A similar kind of progression can be seen in Excerpt (5). The trainer
has just demonstrated how the client should move her thigh and told her to lie
down on her belly with a strap attached to her right leg. The client is now lying
down on a training bench and the trainer has directed her to move her right leg
with the strap. To demonstrate that this movement is difficult in this position, he
asks the client to bend up her left, unstrapped leg in line 1 (gör me de här bene ‘do
(it) with this leg’). In line 3, the adverb så (literally ‘so’ but best translated with ‘and’
in English) initiates a concluding result clause: så märker du att de går no ‘so/and
you’ll see that you sure can make it’.
(5) ’Strength to draw’ (fys:002:1).
01 T: *#Gör me de här bene,
do.imp with this-dem leg-def
Do (it) with this leg,
*T is pointing at C’s left leg
#Figure 3 left
02 *# (0.7)
*C, on her belly, bends her left leg up from the knee.
#Figure 3 right
03 T: så märker du att (.) de går ↑no.
so notice-prs you that it go-prs prt
and you’ll see that you sure can make it.
04 T: [.h *Hi]t ända har man kraft å dra,
Up to this point one has the strength to draw,
*T takes a hold of C’s leg.
05 C: [Jå. ]
Yeah.
Figure 3. Trainer points at client’s left calf (left); Client bends left leg at the knee (right).
This excerpt shows clearly the sequential-temporal emergence of the consecutive

relation. The trainer produces the directive as a first prosodically separate unit (l.
1), but it ends with level intonation that suggests continuation. This is followed by
a pause during which the client complies by carrying out the nominated physical
action (l. 2). Upon compliance, the trainer produces a concluding consecutive
clause (l. 3), which then results in a complex action and clause combination with
the preceding directive clause/action. The verbal presentation is intertwined with
embodied acts. When issuing the directive in line 1, the trainer touches the client’s
calf with a pointing gesture (Figure 3 left), thus clarifying the deictic reference
in the directive (de här bene ‘this leg’). Monitoring the client’s physical compli-
ance, as she bends her leg at the knee, the trainer then matches the account-giving
component with the client’s completion of the task (Figure 3 right). We then get
the following sequential structure:
1. T: Directive inviting action (imperative clause)
2. C: Compliance through physical action
The clause/action combination seems pre-designed, when we take the prosody of
the clauses into account. The directive component ends in the speaker’s middle
range, without final lengthening; this differs from Excerpt (1) where the pitch
contour of the directive is falling and the consecutive clause is produced as an
increment-like expansion (see further 5.4 Summary). Thus, the prosody in (5)
projects a continuation which comes in the account-giving component of the
instruction. In addition, the subsequent consecutive clause is initiated smoothly
with no pitch reset, i.e. as a structurally and prosodically fitted completion of the
two-part instruction. The latter part of the consecutive, de går no ‘you can make it’
has a stylized pitch contour, consisting of a low, high and middle tone, resembling
the stylized figure in Finnish described by Ogden, Hakulinen and Tainio (2004).
The contour can signal that the trainer presents the movement as a routine, pre-
dictable achievement in this activity context where he is the expert; indeed, the
stylization could be heard as signifying expert-to-novice talk.
Both of the combined clauses have the structure of a main clause but they are
sequenced so that they are functionally dependent on each other, standing in an
‘if-then’ relation (cf. the conditional clause combination Om du gör med det här
benet, så märker du att det går nog. ‘If you do it with this leg, then you’ll see that you
can make it’). The consecutive clause indexes its position as a “subsequent” unit
through the adverbial connector and inverted VS order (see also Günthner, this
volume). We could go as far as to say that the consecutive VS clause, then, retro-
constructs the preceding imperative clause as a condition to the consequence, thus
making the imperative clause virtually, and at least pragmatically, subordinate to
the consequence. Such a functional view diverts, of course, from traditional views
of dependency relations in clause combinations.
5.3 Corrective instructions: Trainer intervenes in client’s performance
The instances in this category differ from the two previous categories in that the
trainer produces the instruction as a corrective response to the client’s perfor-
mance during an ongoing exercise (cf. Deppermann, 2015). These responses,
then, are interventions, done by reference to a problem (Reed & Szczepek Reed,
2013: p. 328). In the following Excerpt (6), the client is doing squat jumps,
keeping her hands on her waist. First, she makes a verbal comment on the
challenges related to the exercise (l. 1), and then, she actually loses her balance
(l. 4, Figure 4 left). The trainer responds by providing a bit of instruction (l.
9–10, Figure 4 right).
Figure 4. Client loses balance (left); Trainer shows ‘loose hands’, Client complies (right).
(6) ’Balance’ (vhel:ptr2:002:61).

01 C: *Gud va de här e svårt.
God, this is so difficult.
*C balances, makes squat jumps, alternating the support leg
02 T: Ja:a.
Yeah.
03 (6.6)
04 C: *#Ehh
*C loses the balance
#Figure 4 left
05 (0.4)
06 T: A.
Right.
07 (0.8)
08 T: (De) *kan va lättare
it can.prs be.inf easy-cmp
It can be easier
*T stretches his arms out to the sides
09 ti *#hålla händerna löst
inf hold-inf hand-pl-def loose
to keep the hands loose
*C lets her arms down
#Figure 4 right
10 så du får balansera *me dem lite bättre.
so you get-prs balance.inf with them a.little good-cmp
so you can balance a little better with them.
*T lets his arms down
The trainer directs the client to keep her hands out to the sides instead of having
them on her waist, the verbal format being a suggesting declarative with no person
reference (de kan va lättare ‘it can be easier’). He simultaneously demonstrates
the correct performance by moving his arms out to the sides, and the client com-
plies by moving her hands from her waist out to her sides. The trainer goes on to
provide an account for the instruction (l. 9): changing the position of her hands
will improve the balance. The account is produced at a point when the client has
already changed her physical performance: she has responded immediately to the
trainer’s physical demonstration which is intertwined with the verbal suggestion
in lines 8–9. The sequential structure in this instructing sequence can then be
illustrated in the following way:
1. C: Performs physical exercise
2. T: Directive suggesting action (with an embodied cue)
3. C: Complying physical action (in overlap with 2)
The directive and account-giving components are produced in one go, with no
intervening pauses as the client responds to the directive immediately without
pausing in her exercise. She monitors the emergence of the instruction carefully,
responding through altering her performance as soon as she notices the trainer’s
physical demonstration (l. 9) and thus even before the completion of the verbal di-
rective. The introduction to the instructing turn de kan va lättare ‘it can be easier’
contains the adjective “easier” in the comparative degree and can be heard as
projecting a move to improve the performance. The consecutive clause concludes
the instruction by explicating the client’s problem and the point of the intervening
instruction. It also shifts agency to the client from “it can be easier” in the initial
part to “you can balance a little better” in the concluding part of the instruction.
Excerpt (7) provides another case with corrective instruction that shapes the
form of the client’s performance. In this exercise the client is standing on her
knees, and is supposed to lean with her body backwards but without moving her
thighs backwards. The trainer is standing beside the client on the gym mat, on her
knees as well, monitoring the client’s body position.
(7) ‘Lock the hips’ (vgbg:ptr:03: 9).
01 T: Okej (.) tänk att den ska gå
Okay, imagine that there is
02 som en *linje rakt ner,
a line that goes straight down,
*C bends backwards
03 (0.3)
04 T: så att du- (.) *#Här låser du,
so that you here lock-prs you
so that you- Here your leg gets locked,
*T places her hand on the front of C’s thigh
#Figure 5 left
05 *(0.4)*
*C gazes down and complies and keeps thighs vertical*
06 T: så du [inte går ba- (.) *#så du inte-]
so you not go-prs so you not
so you don’t go ba- (.) so you don’t,
*C complies by not bending thighs
#Figure 5 right
07 C: [Jaha (0.3) så¿ (.) (så¿ )]
I see, (like this?)
08 (.)
09 T: Exakt. (.) Så de e höften vi ska jobba me.
Precisely. So it’s the hip we are working with.
Figure 5. Trainer touches front of client’s thigh (left); Client “locks” her hips (right).
The sequence starts with a directive introduced with the verb tänk ‘think, imagine’
which invites the client to self-reflect on her physical positioning (cf. Excerpt 3
above), i.e. keeping a straight posture. The client is supposed to bend herself back-
wards and starts the movement in overlap, bending backwards from her knees.
The trainer initiates a consecutive clause with the subordinator så att ‘so that’ but
aborts it (l. 4), responding to the client’s incorrect performance. She then issues a
directive in a declarative form, här låser du ‘here your leg gets locked’, placing her
hand on the front of the client’s right thigh simultaneously with the deictic adverb
här ‘here’ (Figure 5 left). The client complies immediately by “locking” (Figure 5
right), and the trainer then initiates two subordinate consecutive så-clauses (l. 4),
which, however, are not completed. Both are negated, which is a way of accounting
for the instruction in the sense that the speaker depicts scenarios to be avoided
(apparently, not going backwards from the knees): så att du inte går ba- [bakåt] (.)
så du inte …‘so that you don’t go ba- [backwards], so you don’t …’ (l. 6).
The client shows an orientation to the directive issued on line 4 both by making
an altered movement backwards and by giving a verbal response in overlap with
the consecutives on line 6. She first says jaha ‘I see’, a minimal response showing a
change in epistemic status (see A. Lindström, 1999: p. 81, 123), and then produces
two så ‘so, this way’ manner adverbials, evoking feedback. These physical and ver-
bal responses display a candidate understanding and are probably the reason why
the trainer leaves the consecutives in line 6 non-completed. The trainer responds
with a confirmation, exakt ‘precisely’ (l. 8), thereby showing an orientation to the
client’s candidate understanding and providing an expert answer. The instructing
sequence is completed in line 9 with a så-initiated clause that in a “we”-form gives
an upshot of the instructional project, så de e höften vi ska jobba me ‘so it’s the hip
we are working with’, as the trainer is simultaneously clapping her hands on her
own hips. This formulation stands in a more independent pragmatic, grammatical
and prosodic relation to the preceding utterances, aiming at consolidating the par-
ticipants’ intersubjective understanding of what the whole instructional project
was about (see Heritage & Watson, 1980 on formulations), and does not directly
belong to the local action pattern [directive] & [account].
To summarize, Excerpt (7) shows the unfolding of an instructing sequence

that begins with the trainer’s intervention in the client’s performance. The client
responds quickly to the directive, which in turn makes the trainer match the pro-
duction of the accounting consecutive clauses with the response and abort them
when the client’s action displays sufficient understanding. This trajectory can be
formalized with the following steps:
1. C: Performs physical exercise
2. T: Directs C to correct performance (declarative with embodied cue)
3. T: Negated account in a consecutive clause (aborted in response to step 4)
4. C: Complying physical and verbal action (in overlap with 3)
5. T: Confirmation of C’s understanding and conclusion of the project
Examples of this kind show how the mechanisms of clause combining and con-
struction of multi-clausal turns are locally sensitive to the participants’ mutual
embodied and verbal actions. In (7) we can see an aborted clause-combining tra-
jectory in which the abortion responds to an action by the recipient that makes
the completion of the account-giving part of the instruction uncalled-for. Such
adjustments of emerging grammatical and interactional structures are rendered
possible by the participants’ focused orientation to their embodied interaction.
5.4 Summary: On projection, expansion and prosody
When considering the temporal, structural and interactional emergence of the in-
structions in our data we can conclude that they materialize through the processes
of projection and expansion (see Auer, 2009; Deppermann & Günthner, 2015;
Hopper & Thompson, 2008; Pekarek Doehler et al., 2015), but not in a uniform
manner. There is variation in how the clause and action complexes are achieved,
from smoothly conjoined bi-clausal units to sequentially emerging products whose
part-clauses and part-actions do not follow one another in a continuous flow of
talk. In order to understand the workings of the actual combinatory practices, we
now turn to a summarizing comparison of the structural and prosodic features of
the excerpts analyzed above.
The most canonical way of producing an instruction in our collection is
through a consecutive clause combination in which the directive and consecutive
components are delivered in a single, coherent intonation unit. An example of this
variant was seen in Excerpt (6), here schematized in Table 5 and further illustrated
with the pitch trace in Figure 6.

(6) De kan va lättare ti hålla så du får balansera me dem
händerna löst lite bättre.
It can be easier to keep the hands so you can balance a little better
loose with them.
300
200
150
f0 (HZ)
100
70
50
(de) kanva lät ta re ti hål la hän de na löst så du får ba lan se ra me dem li te bätt re
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3 3.1
Time (s)
Figure 6. Pitch trace of the directive and consecutive clauses in Excerpt (6); male
speaker.6
This instance shows all the features of classical grammatical clause combining
with an initial main clause (in a declarative form), the conjunction så ‘so’ as the
lexical conjoiner, and a concluding adverbial clause with SV order. The pitch of the
directive clause stays high, with some final lengthening on the word löst ‘loose’),
and the speaker continues the same pitch contour with no gap. The consecutive
clause has a falling contour and the loudness decreases to the degree that the last
syllable is whispered.
Projection, that is, one action foreshadowing another (Auer, 2005), becomes
an issue in cases where the directive component is followed by a pause that seems
to be designed to be there until the complex action is fulfilled with a consecutive
part. In other words, there seems to be “more to come” after the completion of
the initial directive. Excerpt (5), schematized in Table 6, provided an illuminat-
ing instance of prosodic projection, but prosodic projection is also present in
Excerpts (3), (4), (6) and (7).
6. The waveform is not shown in this figure because the sound file contains a good deal of
background noise from the gym. We have used the software Praat (Boersma & Weenink, 2018)
for the acoustic analyses.

(5) Gör me de här bene, så märker du att (.) de går ↑no.
Do (it) with this leg, and you’ll see that you sure can make it.
The pitch contour of the directive component ends in the middle range, without
final lengthening. This is followed by a gap during which the client complies
with the directive, and the trainer monitors her behavior. He then initiates the
subsequent consecutive clause by picking up the prior pitch contour where it left
off, without a pitch reset (see Figure 7). Thus, we have two intonation phrases that
cohere prosodically and in which the latter one completes the intonational trajec-
tory (see Couper-Kuhlen, 2012 for the prosody of clause combining). However,
the latter part of the account (de går no ‘you can make it’) does not end as low as
usually in a terminal declination (cf. Figure 6), which is the result of the slightly
stylized, “curved” patterning of the intonation contour that places stress on the
final modal particle no(g) ‘sure(ly)’ (see the analysis in Section 4.2).
300
200
150
100
f0 (HZ)
gö me de hä be ne (0.7) så mär ker du att (.) de går no
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3 3.1
Time (s)
Figure 7. Pitch trace and waveform of the directive and consecutive clauses in
Excerpt (5); male speaker.
The account-giving components of the instructions are not strongly projected in

structural terms, because the directive component is expressed through declarative
or imperative main clauses that formally could stand on their own. This is com-
parable with causal clause combinations in cases where the causal clause stands
in a subsequent position (see Ford, 1993 on English; Scheutz, 2001 on German),
or when the protasis is postponed in a conditional clause combination (see Auer,
2000 for German). However, there is often pragmatic projection involved. Many
directives, i.e. the initial parts of the instruction pattern, are formulated in a man-
ner that leaves the question of their ‘why that now’ open, which then is accounted
for in the consecutive clause. For example, in Excerpt (5), the directive to move the
leg that is not in the focus of the treatment may call for a motivation. Of course,
in some types of instruction, the understanding of why something is done is
the learnable and may thus be unexpressed to test the other’s competence (see
Deppermann, 2015). We also observed that some forms of projection combine the
structural and pragmatic as, for example, in directives that are constructed with a
presentational format and contain an adjective in the comparative degree (de kan
va lättare ‘it can be easier’ in Excerpt 6): these formal and semantic features induce
an expectation of an explication of what was presented as being improvable.
In some cases the account-giving component comes after a directive that has
the features of an already finished turn. Such consecutives can be understood as
products of structural expansion, such as in our initial example (1); the clause
combination is represented in Table 7.

(1) Sätt ännu eh handflatan ne:, eller Så att den där axeln
liksom upp mot take. °får°
Turn your um palm to face downward, I mean, So that the shoulder gets
upward towards the ceiling. [relaxed]
Here, the directive component ends with falling pitch (see Figure 8), as well as
with other turn-final features, that is, final lengthening, diminishing loudness
and creaky voice (see Huhtamäki, 2012 for final features of questions in Finland
Swedish). It is followed by a pause during which the trainer monitors the client’s
complying action and produces an acceptance token (not too different from
Excerpt 5, although that one does not contain a verbal acknowledgement).
Nonetheless, the consecutive component follows as a glued-on expansion of the
directive turn, rather than as a new beginning: it has no pitch reset in the onset,
and it ends with a fall, and at the very end in a whisper (decreasing volume can be
seen in the waveform in Figure 8).
The prosodic design, then, suggests that the account-giving component was
not a pre-planned part of the instruction, but it is linked to the directive through
lexical and prosodic incrementing practices, which, in the end, result in some-
thing that post hoc could be analyzed as a complex clause and action combination
(for a similar point, see Stoenica & Pekarek Doehler, this volume on incremented
400
300
200
150
100
f0 (Hz)
75
sätt än nu eh hand fla tan ne el ler lik som opp mot ta ke (0.5) jep (0.2) så attdendä ax eln får
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5

Time (s)
Figure 8. Pitch trace and waveform of the directive and consecutive clauses in
Excerpt (1); female speaker.
relative clauses). Such cases, then, represent extensions of clausal dependency

relations beyond the “sentence” (see Mithun, 2008). In support of this analysis,
we can note that there is a relatively strong pragmatic projection also in this case.
The trainer is holding the client’s left leg, stretching it upwards, but the directive
concerns a small detail of hand positioning, the purpose of which, given the focus
on the leg, may appear obscure for the client.
In the data, and in the detailed analyses above, we can see that the emergence
of most directive and account-giving clause/action combinations involves a se-
quence of dialogic verbal and embodied moves. The regularity with which the
combinations [directive & account] appear in the data suggests that they consti-
tute a routine way of speaking, a conventionalized pattern for accomplishing a
complex social action (see Hopper 1998). In this respect we cannot say that the
combinations emerge as results of local, unforeseen interactional contingencies
(see Etelämäki, Couper-Kuhlen & Laury, 2017 for the division-of-labor pattern).
To be sure, there is a pause between the constitutive parts of many instructions,
but the pauses are often designed features of the instructing sequence. As our
stepwise sequence schemas in the above sections suggest, the professionals de-
ploy pauses strategically, leaving space for the client’s compliance or displays of
understanding after a directive has been issued. Having monitored the client’s
performance, sometimes with a concomitant acknowledgement token, the trainer
then concludes the instruction with an account linked to the directive, strengthen-
ing its “pedagogical” or motivational effect. Similarly, trainers can leave pauses
between the directive and account in preparatory instructions as they demonstrate
the nominated action during the pause. This then showcases how grammar can be
fitted to sequences and trajectories of verbal and embodied actions (see Keevallik,
2018). The strength of these designed features varies from cases with prosodically
projected second parts to more ambiguous cases that materialize through expan-
sion and in which some more contingent factors can come into play. In the general
dialogic architecture, however, the clause and action combinations emerge across
participants and modalities: the directive foresees a complying move, the produc-
tion of which in turn “enables” the trainer to deliver the account. Sometimes these
moves can overlap: trainers produce the clause and action combination in one
go or abort the production of a projected account, when they see that the client
already executes the directed action in a satisfactory manner.
6. Conclusion
We have studied how directives are issued by trainers to clients partaking in

physical training, and investigated the ways in which invitations to action, compli-
ances with them, and accounts for them emerge through grammatical, prosodic
and embodied resources. We identified a two-part pattern [directive & account]
that accomplishes a more complex structural and pragmatic unit, here labeled as
instruction. A related kind of combining of actions and clauses, [directive & com-
mitment] or [commitment & directive], has been analyzed by Couper-Kuhlen and
Etelämäki (2014, 2017). The instructions consisting of a directive and an account-
giving component are grammatically formed of consecutive clause combinations
in which the directive most often is a declarative or an imperative clause; these are
linked together with a consecutive conjunction (så ‘so’, så att ‘so that’) or adverb (så
‘so/and’), in the former case through hypotaxis and in the latter through parataxis.
The complex clause and action combinations occur both in preparatory and cor-
rective instructional sequences.
The directive component is often followed by the client’s verbal or physical
compliance, or, indeed, the trainer can design the instruction in a way that leaves a
gap after the directive so that the client has a temporal space to comply or display
understanding, or the trainer can self-demonstrate the directed movement in this
space. The account-giving component, then, comes as an elaboration of the direc-
tive to increase the pedagogical moral of the instruction, giving the clients clues
for what they can achieve through an exercise or a detail in its execution. Because
not only execution but also understanding are the goals of the instructed exercises,
the trainers do not simply exert their “deontic authority” to produce plain direc-
tives (cf. Stevanovic & Peräkylä, 2012).
Both parts of the instruction, the directive and the account, can be produced
in one intonation phrase, but when there is a designed gap between the parts, the
account is projected by a “continuing” intonation at the end of the directive. In

some cases the account emerges as an expansion, as the directive has final intona-
tion. But also in these instances there is a rather strong pragmatic projection: the
directive itself does not inform the client about its purpose and may even seem
unmotivated when considering what has been focused in the ongoing training
activity. We can then conclude that the pattern [directive] & [account] is a conven-
tionalized structural resource that displays variation in some detail, for example,
in that the directive can be a declarative or an imperative.
Our data has targeted the Swedish language, or two national variants of it
which do not display any major differences in the deployment of the combination
[directive & account], disregarding some pragmatic modulations of imperative-
formatted directives in Sweden Swedish (the format tänk på ‘consider this’ to
invite introspection as well as action). The pattern of [directive & account] seems
so fundamentally rooted in an instructing activity context that we have reason to
believe that at least the underlying phenomenon, of combining clauses and ac-
tions to produce complex instructions of the kind we have studied here, is more
universal and to be found in other languages as well.
Transcription symbols7
but emphasis
bu:t lengthening of a sound
°but° sotto voce
(but) parenthesized words are possible hearings
bu- cut-off word or unfinished intonation unit
, final level intonation
. final falling intonation
¿ final slightly rising intonation
↑ locally rising intonation
* timing of embodied action with talk (beginning/end point)
# shows the exact moment for a screen shot (“Figure”)
7. For other standard CA transcription conventions, see Ochs, Schegloff and Thompson
(1996: p. 461–465).
Grammatical glosses
cmp comparative inf infinitive marker or form

dem demonstrative pl plural
def definite prs present tense
imp imperative prt particle
References
Auer, P. (2000). Pre- and postpositioning of wenn-clauses in spoken and written German. In E.
Couper-Kuhlen, & B. Kortmann (Eds.), Cause – condition – concession – contrast: Cognitive
and discourse perspectives (pp. 173–204). Berlin: de Gruyter.
https://doi.org/10.1515/9783110219043.2.173
Auer, P. (2005). Projection in interaction and projection in grammar. Text 25(1), 7–36.
Auer, P. (2009). On-line syntax: thoughts on the temporality of spoken language. Language Sci-
Boersma, P., & Weenink, D. (2018). Praat: Doing phonetics by computer. Version 6.0.39, April
3rd 2018. http://www.fon.hum.uva.nl/praat/
Couper-Kuhlen, E. (2012). Turn continuation and clause combinations. Discourse Processes
49(3/4), 273–299. https://doi.org/10.1080/0163853X.2012.664111
Couper-Kuhlen, E., & Ono, T. (2007). Incrementing in conversation. A comparison of practices
in English, German and Japanese. Pragmatics 17(4), 513–552.
Couper-Kuhlen, E., & Etelämäki, M. (2014). On divisions of labor in request and offer environ-
ments. In P. Drew, & E. Couper Kuhlen (Eds.), Requesting in social interaction (pp. 115–144).
Amsterdam/Philadelphia: Benjamins.
Couper-Kuhlen, E., & Etelämäki, M. (2017). Linking clauses for linking actions: Transforming
requests and offers into joint ventures. In R. Laury, M. Etelämäki, & E. Couper-Kuhlen
(Eds.), Linkings clauses and actions in social interaction (pp. 176–200). Helsinki: SKS.
interaction. Cambridge: Cambridge University Press.
De Stefani, E. & Gazin, A.-D. (2014). Instructional sequences in driving lessons: Mobile partici-
pants and the temporal and sequential organization of actions. Journal of Pragmatics 65,
Deppermann, A. (2015). When recipient design fails: Egocentric turn-design of instructions in
driving school lessons leading to breakdowns of intersubjectivity. Gesprächsforschung 16,
63–101.
Deppermann, A., & Günthner, S. (Eds.). (2015). Temporality in interaction. Amsterdam: Ben-
jamins. https://doi.org/10.1075/slsi.27
Drew, P. & Couper-Kuhlen, E. (Eds.). (2014). Requesting in social interaction. Amsterdam/
Philadelphia: Benjamins. https://doi.org/10.1075/slsi.26
Ervin-Tripp, S. (1976). Is Sybil there? The structure of some American English directives. Lan-
guage in Society 5(1), 25–66. https://doi.org/10.1017/S0047404500006849
Etelämäki, M., Couper-Kuhlen, E., & Laury, R. (2017). Introduction. In R. Laury, M. Etelämäki,
& E. Couper-Kuhlen (Eds.), Linking clauses and actions in social interaction (pp. 11–23).
Helsinki: SKS.
Fisher, D., & Frey, N. (2014). Checking for understanding: Formative assessment techniques for
your classroom. Alexandria, VA: ASCD.
Ford, C. E. (1993). Grammar in interaction: Adverbial clauses in American English conversation.
Garfinkel, H. (2002). Instructions and instructed actions. In H. Garfinkel, & A. Warfield Rawls
(Eds.), Ethnomethodology’s Program, (pp. 197–218). Lanham, MD: Rowman & Littlefield
Publishers.
Goodwin, C. (2000). Action and embodiment within situated human interaction. Journal of
Pragmatics 32, 1489–1522. https://doi.org/10.1016/S0378-2166(99)00096-X
Grahn, I.-L., & Huhtamäki, M. (2019). Frasformade instruktioner under fysisk aktivitet – form
och funktion i interaktionell belysning [Phrase-formed instructions during physical
activity – form and function in an interactional illustration]. In M. Bianchi, D. Håkans-
son, B. Melander, L. Pfister, M. Westman, & C. Östman (Eds.), Svenskans beskrivning 36
(pp. 65–78). Uppsala: Dept. of Scandinavian Languages, Uppsala University.
Henricson, S., & Nelson, M. (2017). Giving and receiving advice in higher education: Compar-
ing Sweden-Swedish and Finland-Swedish supervision meetings. Journal of Pragmatics
Heritage, J., & Watson, D. R. (1980). Aspects of the properties of formulations in natural conver-
sations: Some instances analyzed. Semiotica 30(3/4), 245–262.
Hopper, P. J. (1998). Emergent grammar. In M. Tomasello (Ed.), The new psychology of language:
Cognitive and functional approaches to language structure (pp. 155–175). Mahwah, HJ:
Erlbaum.
Hopper, P., & Thompson, S. A. (2008). Projectability and clause combining in interaction. In R.
Laury (Ed.), Crosslinguistic studies of clause combining: The multifunctionality of conjunc-
tions (pp. 99–123). Amsterdam/Philadelphia: Benjamins.
https://doi.org/10.1075/tsl.80.06hop
Huhtamäki, M. (2012). Prosodiska mönster hos frågor: en undersökning av Helsingforssvenska
samtal [Prosodic patterns in questions: A study of Helsinki Swedish conversations]. Språk
och stil 22(2), 153–184.
on Language and Social Interaction 46(1), 1–21.
https://doi.org/10.1080/08351813.2013.753710
Keevallik, L. (2015). Coordinating the temporalities of talk and dance. In A. Deppermann, & S.
Günthner (Eds.), Temporality in interaction (pp. 309–336). Amsterdam: Benjamins.
Keevallik, L. (2018). What does embodied interaction tell us about grammar? Research on Lan-
guage and Social Interaction 51(1), 1–21. https://doi.org/10.1080/08351813.2018.1413887
Limberg, H., & Locher, M. A. (Eds.). (2012). Advice in discourse. Amsterdam/Philadelphia:
Benjamins. https://doi.org/10.1075/pbns.221
Lindström, A. (1999). Language as social action: Grammar, prosody, and interaction in Swedish
conversation. (Diss.) Uppsala: Uppsala University.
Lindström, J. (2014a). On the place of turn and sequence in grammar: Verb-first clausal con-
structions in Swedish talk-in-interaction. Pragmatics 24(3), 507–532.
https://doi.org/10.1075/prag.24.3.04lin
Lindström, J. (2014b). Front field negation in spoken Swedish: A regional archaism? In I. Ta-
avitsainen, A. H. Jucker, & J. Tuominen (Eds.), Diachronic corpus pragmatics (pp. 237–253).
Amsterdam: Benjamins. https://doi.org/10.1075/pbns.243.14lin
Lindström, J., Lindholm, C., Norrby, C., Wide, C., & Nilsson, J. (2017). Imperatives in Swedish
medical consultations. In M.-L. Sorjonen, L. Raevaara, & E. Couper-Kuhlen (Eds.), Impera-
tive turns at talk (pp. 299–324). Amsterdam/Philadelphia: Benjamins.
Lindwall, O. (2008). Lab work in science education: Instruction, inscription, and the practical
achievement of understanding. (Diss.) Linköping: Linköping University.
Lindwall, O., & Lymer, G. (2014). Inquiries of the body: Novice questions and the instructable
observability of endodontic scenes. Discourse Studies 16(2), 271–294.
https://doi.org/10.1177/1461445613514672
Matthiessen, C., & Thompson, S. A. (1988). The structure of discourse and ‘subordination’.
In J. Haiman, & S. A. Thompson (Eds.), Clause combining in grammar and discourse
(pp. 275–329). Amsterdam. Benjamins. https://doi.org/10.1075/tsl.18.12mat
Mithun, M. (2008). The extension of dependency beyond the sentence. Language 84(1), 69–119.
https://doi.org/10.1353/lan.2008.0054
Mondada, L. (2009). Emergent focused interactions in public places: A systematic analysis of
the multimodal achievement of a common interactional space. Journal of Pragmatics 41,
Norrby, C., Wide, C., Lindström, J., & Nilsson, J. (2012). Finland-Swedish as a non-dominant
variety of Swedish – Extending the scope to pragmatic and interactional aspects. In R. Muhr
(Ed.), Non-dominant varieties of pluricentric languages: Getting the picture (pp. 49–60).
Frankfurt am Main: Peter Lang.
Norrby, C., Wide, C., Lindström, J., & Nilsson, J. (2015). Interpersonal relationships in medical
consultations: Comparing Sweden-Swedish and Finland-Swedish address practices. Jour-
nal of Pragmatics 84, 121–138. https://doi.org/10.1016/j.pragma.2015.05.006
Ogden, R., Hakulinen, A., & Tainio, L. (2004). Indexing ‘no news’ with stylization in Finnish.
In E. Couper-Kuhlen, & C. E. Ford (Eds.), Sound patterns in interaction: Cross-linguistic
studies from conversation (pp. 299–334). Amsterdam/Philadelphia: Benjamins.
https://doi.org/10.1075/tsl.62.16ogd
Pekarek Doehler, S., De Stefani, E., & Horlacher, A.-S. (2015). Time and emergence in grammar:
Dislocation, topicalization and hanging topics in French talk-in-interaction. Amsterdam/
Philadelphia: Benjamins. https://doi.org/10.1075/slsi.28
Reed, D., & Szczepek Reed, B. (2013). Building an instructional project. Actions as components
of music masterclasses. In B. Szczepek Reed, & G. Raymond (Eds.), Units of talk – units of
action (pp. 313–341). Amsterdam/ Philadelphia: Benjamins.
https://doi.org/10.1075/slsi.25.10ree
Reuter, M. (1992). Swedish as a pluricentric language. In M. Clyne (Ed.), Pluricentric languages
(pp. 101–116). Berlin: Mouton de Gruyter.
Scheutz, H. (2001). On causal clause combining: The case of ‘weil’ in spoken German. In M.
Selting, & E. Couper-Kuhlen (Eds.), Studies in interactional linguistics (pp. 111–140). Am-
sterdam: Benjamins. https://doi.org/10.1075/sidag.10.07sch
Sorjonen, M-L., Raevaara, L., & Couper-Kuhlen, E. (Eds.). (2017). Imperative turns at talk.
Amsterdam/Philadelphia: Benjamins. https://doi.org/10.1075/slsi.30
Stevanovic, M., & Peräkylä, A. (2012). Deontic authority in interaction: The right to announce,
propose, and decide. Research on Language and Social Interaction 45(3), 297–321.
https://doi.org/10.1080/08351813.2012.699260
Szczepek Reed, B., Reed, D., & Haddon, E. (2013). NOW or NOT NOW: Coordinating restarts
in the pursuit of learnables in vocal masterclasses. Research on Language and Social Interac-
tion 46(1), 22–46. https://doi.org/10.1080/08351813.2013.753714
Wide, C., & Lyngfelt, B. (Eds.). (2009). Konstruktioner i finlandssvensk syntax: skriftspråk, sam-
tal, dialekter. [Constructions in Finland-Swedish syntax: Written language, conversation,
dialects.] Helsingfors: Svenska litteratursällskapet i Finland.
Chapter 10
Right-dislocated complement clauses

in German talk-in-interaction
(Re-)specifying propositional referents
of the demonstrative pronoun das
Nadine Proske and Arnulf Deppermann
This contribution deals with right-dislocated complement clauses with the

subordinating conjunction dass (‘that’) in German talk-in-interaction. The
bi-clausal construction we analyze is as follows: The first clause, in which one
argument is realized by the demonstrative pronoun das (‘this/that’), is syntacti-
cally and semantically complete; the reference of the pronoun is (re-)specified
by adding a dass-complement clause after a point of possible completion (e.g.,
aber das hab ich nich MITbekommen. (0.32) dass es da so YOUtubevideos gab.
(‘But I wasn’t aware of that. That there were videos about that on YouTube.’).
The first clause always performs a backward-oriented action (e.g., an assess-
ment) and the second clause (re-)specifies the propositional reference of the
demonstrative, allowing for a (strategic) perspective shift. Based on a collection
of 93 cases from everyday conversations and institutional interactions, we found
that the construction is used close to the turn-beginning for referring to and
(re-)specifying (parts of) another speaker’s prior turn; turn-internal uses tie
together parts of a speaker’s multi-unit turn. The construction thus facilitates an
incremental constitution of meaning and reference.
Keywords: right-dislocation, complementizer, German, demonstrative
1. Introduction
This contribution1 deals with a bi-clausal construction in spoken German that

has not been empirically investigated to date. It consists of a main clause with a
1. We thank the editors and Elwys De Stefani for their comments on earlier versions of this
paper. We also thank our colleagues Julia Kaiser, Thomas Spranz-Fogasy, and Jörg Zinken for
helpful discussions. Thanks to Allison Adelman for checking our English.
https://doi.org/10.1075/slsi.32.10pro
276 Nadine Proske and Arnulf Deppermann
demonstrative pronoun (das ‘that’) and a complement clause with the complemen-
tizer dass (‘that’), as in the following example: aber das hab ich nich MITbekommen.
(0.32) dass es da so YOUtubevideos gab. (‘But I wasn’t aware of that. That there
were videos about that on YouTube.’). In the literature, the demonstrative in the
first clause is seen as co-referential with the ‘right-dislocated’ complement clause,
that is, it is seen as a cataphoric device (see Section 2). Yet, the literature stresses
that the neuter demonstrative das can have various uses, which include anaphoric,
cataphoric, anadeictic, and catadeictic reference (Ahrenholz, 2007). From an
online-perspective on talk-in-interaction, this raises the question of how the
pronoun reference within a right dislocation construction is established: Upon
production of the first clause, the recipient cannot know that the speaker will add
a complement clause. Sometimes it is even obvious (because of pauses, hesitations,
and prosodic features) that the speaker has not planned the complement clause in
advance. Therefore, the demonstrative cannot be interpreted as cataphoric upon its
occurrence; rather, it must first be interpreted like any other demonstrative. Once
the second clause is produced, though, the interpretation of the pronoun must be
adjusted according to the content of the complement clause and the first clause
becomes retrospectively interpretable as a matrix clause. The right dislocation
construction we discuss in this article is thus intrinsically emergent, because the
reanalysis of the anaphoric demonstrative as a cataphoric pronoun occurs online,
that is, while its production and reception are in progress. Points of completion on
different levels of linguistic description (syntax, prosody, semantics, pragmatics)
contribute to the interpretation of the pronoun and the right dislocation construc-
tion as a whole. The construction’s conventionalized status makes it possible to
use it locally in an incremental fashion by expanding a turn after a clear point of
completion or by using prosodic projection at the end of the first clause to signal
more upcoming talk.
Our study focuses on the ways in which the bi-clausal right dislocation con-
struction is used to accomplish social actions. We will show that the first clause
performs a backward-oriented action (such as accepting a proposal, assessing
another participant’s action, or asking a question which builds on the prior turn),
while the second clause has a more forward-oriented function. By specifying or
even shifting the reference of an argument of the verb of the first clause, a speaker
can move on to a different topic, or focus on a different facet of a current topic.
Thus, the analysis of the complement clause importantly focuses on semantic and
referential practices.
Even though the right dislocation of complement clauses is not seen as be-
longing to Standard German in normative grammars, our data from spontaneous
talk-in-interaction show that it is not rare at all, but is a routinely used, productive
bi-clausal construction. We argue that the construction is well motivated by the
Chapter 10. Right-dislocated complement clauses in German talk-in-interaction 277
contingencies of talk-in-interaction: It allows for an incremental specification of

action and meaning over the course of its production. Reasons for specification
may be a co-participant signaling a lack of understanding, the speaker’s emergent
online planning, the identification of ambiguities or vagueness via self-monitoring
of ongoing turn production, or even rhetorical uses to shift the topic or to replace
referents of predications. The construction offers a smooth, conventionalized
way to expand a turn.
2. Right dislocation in German
2.1 General properties of the construction [[NP V dasi] [dass NP VP]i]
The construction consists of two clauses: A main clause containing the neuter de-
monstrative pronoun das2 (either subject or object of the verb) and a complement
clause with the complementizer dass that is co-referential with the demonstrative.
The first clause of the construction could in principle be used independently as it
is syntactically and semantically complete;3 the pronoun das anaphorically refers
to an antecedent in the ongoing verbal interaction (or to something inferable
from it)4 or deictically to something that is not verbally expressed but present
in the situation.
Clauses with demonstrative pronouns are very frequent in spoken German;5
the demonstrative is mostly not specified any further. If speakers want to specify
the reference of das, however, they can make use of an expansion, which can take
the form of a complement clause. If the dass-clause re-specifies rather than merely
explicating the reference of das, this can lead to a reinterpretation of the matrix
clause. The occurrence of a following complement clause that semantically fills
an argument slot of the verb of the main clause, which was already occupied by
2. The articulation of the demonstrative varies according to regional dialect and other factors
such as register (das, des, dis, dit, dat).
3. The matrix clause is always semantically complete in the sense that it contains a full proposi-
tion, that is, all obligatory semantic roles of the verb are realized. It may, however, not be suf-
ficiently specified if the reference of the demonstrative is unclear or ambiguous.
4. In line with Consten and Schwarz-Friesel (2007), we use anaphoric not only for references
to antecedents, that is, previously mentioned referents and propositions, but also to concep-
tual entities that have to be inferred (e.g., by metonymic relations) from anchors in prior talk
(“indirect anaphora”).
5. See e.g., Proske (2013) for the frequency and functions of copular clauses with das as a sub-
ject, which are one of the most frequent formats that are often expanded by complement clauses.
the demonstrative, retrospectively turns the main clause into a matrix clause. The
resulting construction is treated as a type of right dislocation in the literature. We
briefly illustrate its structure with an example. Afterwards, we describe different
prosodic realizations and discuss how they impinge on the projective properties
of the first clause.
In (1), two friends are talking about a TV series. FR has just started watching
the third season, while EG has already seen it when it was originally broadcast.
FR tells EG that apparently there was one particularly spectacular episode in the
third season, and that people even filmed their reactions to this episode and put
them on YouTube (l. 1–6). EG reacts with a multi-unit turn containing several
clauses with anaphora (l. 7–12),6 the last one of which is aber das hab ich nich
MITbekommen (‘But I wasn’t aware of that.’, l. 14).
(1) Telephone conversation among friends [FOLK_E_00084_T_01, c579]7
01 FR °h ich k weiß nur dass da damals so ganz viele
((omission, approx. one second))
02 reakTIOnen–
03 °h ähm geZEIGT worden sind;
‘I only know that at the time many reactions, um, were shown,’
04 auch auf so: YOUtubevideos–
‘also in such YouTube videos,’
05 °h wie leute ähm irgendwie total stark auf diese EIne FOLge reagiern.
‘of how people somehow reacted very strongly to this one episode.’
06 es gibt wohl so eine gAnz krasse [FOLge. ]
‘There must be one extremely hard [episode.’]
07 EG [jA das kann] SEIN.
[‘Yeah, that’s] possible.’
08 (0.64)
09 EG also mir würde jetz da was EINfallen <<:-)> auf jeden fall.>
‘I mean, I can think of something now for sure.’
10 (0.61)
11 FR ok[ay. ]
12 EG [kann sch]on SEIN.
‘It’s possible.’
6. In a more detailed analysis that cannot be given here, these other anaphora help disambigu-
ate the reference of das in line 14, as some of the predications are contradictory (einfallen vs.
nicht mitbekommen).
7. The transcription of the examples follows the conventions of GAT 2, as outlined by Selting
et al. (2009). Each line represents one intonation unit. There are five different terminal pitch
contours: low falling (.), slightly falling (;), level (–), slightly rising (,), and high rising (?).
Primary accents are marked by upper case of the whole accented syllable and secondary accents
by upper case of the vowel of an accented syllable.
13 (0.42)
14 EG aber das hab ich nich MITbekommen.
but 3sg.n.acc aux.1sg.prs 1sg.nom not hear.inf
‘But I wasn’t aware of that.’
15 (0.32)
16 dass es da [so YOUtubevideos ] gab.
comp 3sg.n.nom there such youtube.video.pl.acc give.3sg.pst
‘That there were [videos about that on YouTube.]’
17 FR [okay das WUSStest du n]
[‘Okay, you didn’t know that.’]
18 EG [nee. ]
[‘No.’]
19 FR [ah ja das] is auch schon lange HER.
[‘Ah well, that] is long ago anyway.’
The pronoun das in line 14 can be interpreted as a “complex anaphor” (Consten,

Knees, & Schwarz-Friesel, 2007), which can refer to FR’s prior turn as a whole
or just to a part of it. Thus it could mean that EG was ‘not aware’ either of all of
what FR has said in lines 1–6 or of only one of the facts she mentioned. FR’s lack
of uptake after the clause in line 14 is taken by EG to indicate that a specifica-
tion (l. 16) is necessary. EG then adds a dass-complement clause, dass es da so
YOUtubevideos gab. (‘That there were videos about that on YouTube’, l. 16), which
specifies the reference of das. It narrows down the issues that EG was not aware of
to a subset of those mentioned in FR’s turn. Only when EG has started to give the
specification does FR display that she understands what EG is referring to (l. 17).
While in this example the interactive motivation for specifying the pronominal
reference by adding a complement clause seems to be transparent, in most cases
in our collection we do not find that speakers add the complement clause because
of (the lack of) listeners’ responses. Yet, we find syntactic, semantic, pragmatic,
and prosodic indicators that index the possible (in-)completeness of the preceding
clause and thus project turn continuation.
Literature on the grammar of German (Altmann, 1981: 167ff.; Zifonun,
Hoffmann, & Strecker, 1997: 548) claims that the matrix clause of the right dis-
location construction is always prosodically independent, separated by a pause,
and concluded with a falling final intonation contour as in Excerpt (1). Yet, our
data show that the complement clause and the matrix clause are almost as often
produced together in the same intonation unit as they are produced as separate
intonation units. If the matrix clause constitutes an intonation unit of its own, it
can be produced with a falling, level, or rising contour.8
We consider cases such as (1), in which the first clause is complete on all
levels of linguistic description (syntax, semantics, pragmatics, and prosody)
and therefore ends in a “complex TRP” (Ford & Thompson, 1996), to be clear
cases of expansion (or: turn continuation, see Auer 1996; Couper-Kuhlen & Ono,
2007; Ono & Thompson, 2012).9 In contrast, cases in which the first clause is not
complete on all levels are treated here as projecting (on at least one level, cf. Auer,
2005, 2010), that is, the first clause makes expectable a continuation, although
not necessarily one by a complement clause. Thus, projection can result from
syntactic, semantic, pragmatic, or prosodic incompleteness. As the construction
under scrutiny contains a syntactically and semantically complete first clause in
most cases,10 it is prosody and pragmatics that can be seen as decisive regarding
projection here. If the first clause does not end in a point of prosodic comple-
tion, this still does not mean that the form of the continuation is pre-defined in
any way (the continuation may, for example, also take the form of a simple main
clause). The prosodically projecting cases can be split into two groups: a) the two
clauses are produced in one intonation unit (mostly with two focal accents, one
per clause) (see Excerpt (2)), or b) each clause is produced as a separate intonation
unit and the first one ends in a non-final (rising or, more rarely, level) contour (see
Excerpt (3)).11 In the expansion cases, the first clause ends with a final, falling or
high rising, intonation contour and the complement clause is added as a separate
intonation unit (see Excerpt (1)).
8. 56 of the 93 cases (60%) are produced in two separate intonation units. Of these, 28 have a
falling final contour on the first clause, 19 a rising one (only five of which are high rising), and 8
a level one. 37 of the 93 cases (40%) are produced within one intonation unit; 29 of these intona-
tion units feature two focal accents; 8 have only one focal accent. In one case, the intonation unit
is cut off after the complementizer.
9. This roughly corresponds to what Schegloff (1996) has called increment, a term that we
avoid because of certain ambiguities associated with it (see Deppermann & Proske, 2015 for a
discussion).
10. In some cases, the first clause cannot be regarded as semantically complete, as there is no
antecedent or anchor for the demonstrative. In these cases, it functions as a non-referential
expletive as es in the extraposition construction does. This lack of referential specification can
be regarded as semantic incompleteness.
11. We use the term intonation unit as it is understood in interactional studies on spoken English
and German, that is, as “a stretch of speech uttered under a single coherent intonation contour”
(Du Bois et al., 1993: 47). An intonation unit has at least one focal accent and a terminal pitch
contour following it; among the optional ‘boundary cues’ are features such as audible in-breath,
pause, change in pitch, and change in tempo (see e.g., Chafe, 1994; Selting et al., 2009).
(2) Board game [FOLK_E_00010_T_01, c468]

01 NK eigentlich hab ich mir das NICH denken können
actually aux.1sg.prs 1sg.nom refl.1sg 3sg.n.acc not think.inf can.inf
dass da AUCH noch eine drunter is.
comp there too another_one under.it be.3sg.prs
‘Actually I couldn’t imagine that there is another one under it.’
(3) Mediation meeting (Stuttgart 21) [FOLK_E_00064_T_06, c70]

01 GS eine WERbebroschüre–
02 °h die vielen w wahlbürgern auch suggeRIERT–
03 (.) sie hätten einen NUTzen davon;
04 verKEHRlich einen nutzen;
‘A flyer that suggests to many voters that they will
profit from it traffic-wise’
05 und das war die entscheidende aussage
and 3sg.n.nom be.3sg.past det.f.nom decisive.f.nom statement.nom
die w wir von herrn kefer HÖren (.) wollten,
rel.3sg.f.acc 1pl.nom from mister.dat Kefer hear.inf want.2pl.past
‘and that was the decisive statement that we wanted to hear from
Mr. Kefer’
06 °h dass dieser gar nich geGEben is.
comp 3sg.m.nom at.all not give.ptcp be.3sg.prs
‘that this [profit] is not to be expected at all.’
Pragmatic projection comes into play in addition to prosodic projection. It may

be the case that prosody signals completion in line with syntax and semantics, but
pragmatically the turn cannot be regarded as complete (see Excerpt (4) below).
2.2 Right dislocation in the literature
The right dislocation construction is usually defined as consisting of a clause con-

taining a pro-form and a phrase realized after (‘to the right of ’) the clause that is
coreferential with the pro-form (see e.g., Altmann, 1981: 54f.; Averintseva-Klisch,
2009: 3). The use of the term right dislocation (in German: Rechtsversetzung) was
originally restricted to clause-final phrases with a nominal core, that is, NPs and
PPs (e.g., °h also die fand ich sehr SCHWER, diese PROsabände– ‘well, I found
them very hard [to understand], these prose volumes’ FOLK_E_00059_ T_01,
c742).12 The interactional functions of the construction, most prominently repair,
12. The term was coined by Ross (1967). In the generative framework, the complement clause is
seen as ‘right-dislocated’ from its original position within the matrix clause, which is occupied
by the expletive pronoun instead. Although this transformational view of the constructions
does not adequately capture their emergence in spoken language, we use the term because it
have been described for German in studies on various types of phrasal expan-
sions of clauses (e.g., Auer, 1991, 1996, 2007; Uhmann, 1993).13 The possibility
of dislocating complement clauses to the right is only rarely considered (see e.g.,
Altmann, 1981: 167ff.; Zifonun, Hoffmann, & Strecker 1997: 1475f. for short men-
tions). There are no major studies on the construction in German, neither from a
theoretical syntactic nor from a usage-based or interactional point of view.14 The
construction has two formal relatives that are considered standard in German:15 a)
a matrix clause with one argument realized as a complement clause, and b) a matrix
clause within which one argument is realized as an expletive pronoun (es ‘it’) before
the same argument is realized again as a complement clause (‘extraposition’).16
All three constructions are restricted to matrix clauses with complement-taking
predicates, that is, verbs that take a propositional argument that can be realized
either as a complement clause or a neuter pronoun (es or das). This is also why
right-dislocated clauses have much less in common with right-dislocated NPs
and PPs than with the above-mentioned bi-clausal structures: Right dislocation of
phrases can occur with any verb.
Because of their close structural similarities, literature on matrix clause con-
structions and the extraposition construction is potentially relevant for describing
the functions of the right dislocation construction. Matrix clauses in German and
English have been investigated for their framing functions: complement-taking
verbs often have epistemic, evidential, or evaluative semantics. The framing func-
tion and the projection of upcoming talk has also been the focus of studies on
extraposition. Although the matrix clause is formally complete (in contrast to a
is common in non-transformational approaches (e.g., Altmann, 1981: 54) and in interactional

linguistics (e.g., Pekarek Dohler, De Stefani, & Horlacher, 2015) as well.
13. See also Horlacher (2015) and Pekarek Doehler et al. (2015) for right-dislocated NPs in
spoken French. These authors emphasize that—apart from the function of referential repair,
which the literature on right dislocation in many languages has focused on (see e.g., Geluykens,
1987)—phrasal right dislocation has several functions concerning turn-taking and turn con-
struction, such as dealing with issues of recipiency, as was also found in our example of clausal
right dislocation in Excerpt (1).
14. But see Keevallik’s (2011) investigation of clause combinations with projecting pro-forms
in spoken Estonian.
15. Es is held to be the default expletive pronoun (Gallmann 2009: 822f., 1055), while das is
regarded as a non-standard expletive.
16. If our Excerpt (1) would have been realized as a simple matrix clause construction or
as an extraposition construction, it would look as follows: Aber ich hab (es) nicht mitbekom-
men, dass es da so Youtube-Videos gab. The propositional meaning is the same as that of the
right-dislocated case.
matrix clause without an expletive) and makes an epistemic, evidential, or evalu-

ative statement, it is semantically incomplete because one argument of the verb
has been only formally realized; it projects a propositional referent for which the
pre-posed assertion holds (Couper-Kuhlen & Thompson, 2006, 2008; Günthner,
2009).17 This relationship can also be found for cases of right dislocation in which
the first clause is prosodically projecting but the demonstrative pronoun is non-
referential or cannot be interpreted unambiguously.
3. Data
Our data come from the ‘research and teaching corpus of spoken German’ (FOLK;
see Schmidt, 2014), a growing reference corpus of German talk-in-interaction.
The corpus comprised 133 hours of spoken interactions when we set up our data-
collection.18 It contains a broad variety of private, institutional, and public interac-
tion types (e.g., dinner table conversations, private telephone conversations, work
meetings, classroom interaction, university vivas, panel discussions) with speakers
of German as a first language. We selected 13 interactions (27 hours), for which
all occurrences of dass were inspected manually.19 We found 93 cases of right-
dislocated dass-complement clauses (this means that the construction occurs
approximately every 17 minutes). These 93 cases were analyzed sequentially and
for several formal and functional features.
17. Because of the broad functional range of das, the right dislocation of complement clauses
seems to be much more common in German than in English. The English extraposition con-
struction fulfills more functions because it is used anaphorically more often than German es
(cf. Couper-Kuhlen & Thompson, 2006, 2008, who do not use the term ‘anaphoric’, but describe
the it in the first clause of the English extraposition construction as often both ‘forward-’ and
‘backward-oriented’). The close relation between extraposition and right dislocation in German
has not been discussed in detail in the literature. Moreover, it has not yet been investigated em-
pirically whether it is not es—as claimed in grammars—but das that is the default pronoun ‘re-
placing’ complement clauses realized ‘to the right’ of a matrix clause, at least in spoken German.
18. We built the collection in May 2015. The corpus is available at http://dgd.ids-mannheim.de.
All excerpts can be retrieved by the interaction and contribution numbers in the header above
each excerpt.
19. The corpus contains more than 8,000 occurrences of dass-clauses, which can be retrieved
automatically. For our analysis, we selected interactions from different interaction types, in
order to cover a wide range of usages of the construction.
4. Uses of the right dislocation construction
The functions of the right dislocation construction20 do not differ systematically

depending on the sequential or turn-internal position in which it is used. Rather,
its basic characteristics (bi-partite structure, (potentially) anaphoric demonstra-
tive that is (potentially) co-referential with a complement clause) bear different
weight in different contexts of use. First, we look at (mostly) turn-initial uses, by
which a speaker takes up (parts of) a prior speaker’s turn with the demonstrative.
In these cases, the complement clause specifies a propositional referent that was
introduced by another speaker. Second, we deal with turn-internal uses, by which
a speaker ties together parts of their own multi-unit turn,21 that is, the comple-
ment clause specifies a propositional referent that the speaker has introduced
themselves. Third, we analyze co-constructed cases—other-initiated expansions
and collaborative constructions. In these cases, the reference of the demonstrative
is observably negotiated by the co-participants.
4.1 Co-reference with (parts of) a prior turn
The first clause of a right dislocation construction with which a speaker takes up
(parts of) another speaker’s turn occurs close to the beginning of the turn in most
cases. Mostly, it is turn-initial or the first clause in a turn, occurring, for example,
after a turn-initial particle.22 In the majority of these cases, das refers to a non-
adjacent TCU in the prior speaker’s turn. Reference to the immediately preceding
TCU of the prior speaker’s turn also occurs.23 Ambiguity of the antecedent which
is referred to by das in the first clause, or inability to resolve its reference because
of the distance of the antecedent, can thus be only one reason why referential (re-)
specifications by complement clauses occur.
The following examples show that what later becomes interpretable as the ma-
trix clause of a right-dislocated complement clause typically implements an action
that is backward-oriented (i.e., operates on prior talk), such as asking a question
about the current discourse topic or accepting a proposal. As the pronoun das
20. We refer to the bi-clausal right dislocation construction in an abbreviated way as the ‘right
dislocation construction’.
21. In 37 of the 93 cases the demonstrative is co-referential with (parts of) a prior turn of
another speaker; in 35 it is co-referential with a prior part of the same speaker’s turn; in 21 it is
non-referential or cataphoric.
22. 26 of the 37 cases referring to another’s turn occur close to the turn beginning.
23. 15 of the 37 cases referring to another’s turn refer to an immediately preceding TCU.
refers to something that is treated as given, the verb of the clause (and potentially
additional lexical arguments) gains prominence as new information. Mostly, the
matrix clause is short. The complement clause contains a (re-)specification of the
referent that was already indexed as given.
Excerpt (4) comes from a lesson at a professional school. The teacher (GS) is
talking about the possibility of employees getting support from their boss when
taking part in a training course (l. 1–13).
(4) Lesson at a professional school [FOLK_E_00004_ T_02, c463]
01 GS er unnerstützt es eventuELL–
‘He may support it,’
02 wenn er sacht er stellt_s FREI,
‘if he says he releases [someone from work for a training course]’
03 (0.45) o:der vielLEICHT sogar er unnerstützt es hier finanZIELL,
‘or maybe he would even support it financially.’
04 (0.36) aber im prinZIP könnt er ach sache ich–
‘But in principle he could as well say: I,’
05 also mit ihrem WISsensstand,
‘I mean, with your state of knowledge’
06 (0.36) ich kann sie in der firma ni EIsetze.
‘I cannot utilize you in the company.’
07 (0.44)
08 GS da is kein PLATZ mehr.
‘There is no room anymore.’
09 (0.37)
10 GS ja?
‘Right?’
11 (1.41)
13 GS da machen wir irgendwie n qualifiZIErungsmaßnahme oder SONSCHTwas;
‘You have to take a qualification course or something.’
14 bitte?
‘Yes, please?’
15 (0.7)
16 MB ja;
‘Well,’
17 n hab ich das <<↑> RICHtig jetz verstanden;>
aux.1sg.prs 1sg.nom 3sg.n.acc right now understand.ptcp
‘did I understand this correctly?’
18 dass der chEf sagen könnte JA–
comp det.m.nom boss.nom say.inf can.sbjv.3sg yes
19 (0.23) sie machen die lehrgänge priVAT–
3pl.nom do.prs.3pl det.M.pl.acc training.course.pl.acc private
20 (0.68) sonst könn wir sie: (0.37) nich
otherwise can.prs.1pl 1pl.nom 3pl.acc not
mehr weider (0.22) TRAgen?

more further carry.inf
‘that the boss could say: well, you have to do the training courses
privately, otherwise we cannot keep you any more?’
21 GS wenn des irgendwelche SCHEIne sind;
‘If it concerns some kind of certificate,’
22 (0.2) wie zum bEIspiel wenn
‘as for example if’
23 ich ich sag jetz mal n schweiß
24 (0.24) SCHWEIßfachmann;=ja?
‘let’s say a welder, right?’
25 (0.2) zertifiKAte–
26 die er sacht sie MÜSse_s;
‘certificates that he says you must [have],’
27 dAnn kann er dis NIT verlangen.
‘then he cannot demand it.’
MB starts asking a question in line 17: hab ich das RICHtig jetz verstanden; ‘Did I
understand this correctly?’. This clause ends in a possible syntactic and prosodic
point of completion and can be regarded as semantically complete if das is taken to
refer to GS’s prior turn as a whole. Moreover, as a question, the clause performs a
sequentially first action that creates conditional relevance. However, it does not end
in a pragmatic point of completion, as GS cannot answer the question of whether
MB has understood her correctly unless MB offers a candidate understanding,
because GS does not have access to his mental state. Thus, it is only after MB has
explicated his understanding of GS’s turn with the complement clause in lines
18–20 that GS provides an answer (l. 21–27). Whereas in Excerpt (4), the necessity
for specifying what das refers to depends on the semantics of the matrix clause
verb, there are other cases, in which it arises from contextual factors, such as sev-
eral possible antecedents, as in Excerpt (1) above. Specification is more important
for questions than for other actions, for example responsive epistemic statements
as in (1), as they make relevant a next action by the recipient. In any case, the right
dislocation construction offers a way of ‘chunking’ a turn, that is, of first perform-
ing the main action such as asking a question and then specifying reference.24
Excerpt (5) comes from the same telephone conversation as (1), among two
students. They are talking about their work for a newspaper. FR has committed to
writing an article, but has not done so yet; EG proposes that FR should write an
e-mail to the chief editor explaining that she will still write the article (l. 1–4), so
that he won’t assign the topic to anyone else (l. 5–6).
24. Each clause can be seen as a chunk, thus the backward-oriented action (matrix clause) and
the specification (complement clause) are examples of grammatical and pragmatic “packaging”
(Chafe, 1976).
(5) Telephone conversation among friends [FOLK_E_00084_ T_02, c92]

01 EG ja aber dann sach
‘Yeah, but then tell
02 dann SACH ihm doch einfach du MACHST das dann noch,
then simply tell him you’ll still do it.’
03 und äh JA.
‘And um, well.’
04 (0.35) dann SCHEIB ihm das da halt mit REIN.
‘Then write that into [your e-mail to him].’
05 (0.27) dann MACHT er das auch nich;
‘Then he won’t do that.’
06 °h dann GIBT er das auch keinem andern.
‘Then he won’t give it [your topic] to anyone else.’
07 (0.7)
08 FR ja JA.
‘Yes, of course.’
09 ich schreib ihm
‘I’ll write him’
10 ja das WILL ich auch schreiben.
yes 3sg.n.acc want.1sg.prs 1sg.nom modp write.inf
‘Yeah, that’s what I want to write.’
11 dass ich einfach noch keine ZEIT hatte–
comp 1sg.nom simply yet det.indef.neg.f.acc time.acc have.pst.1sg
‘that I simply had no time [to do it] yet.’
12 aber dass
‘but that’
13 °h ich will halt schreiben dass ich leider noch keine ZEIT hab hatte,
‘I want to write that I regrettably haven’t had any time yet,’
14 aber falls es in ORDnung is,
‘but if it is okay,’
15 dann werde ich das auf jeden fall ende februar in ANgriff nehmen.
‘I’ll start working on it by the end of February.’
FR starts her response with a response token (ja JA., l. 8, and again ja in l. 10)
(Barth-Weingarten, 2011; Golato & Fagyal, 2008) and accepts EG’s proposal with
the clause das WILL ich auch schreiben. (‘Yeah, that’s what I want to write’, l. 10).
The pronoun das refers to a non-adjacent part of EG’s turn in line 2 (du MACHST
das dann noch, ‘you will still do it’). FR demonstrates epistemic autonomy as she
identifies the proposal as her own prior intention by using the modal verb wollen
‘want’. The clause is complete syntactically, semantically, and pragmatically and
ends with a falling intonation contour. After this possible point of completion, FR
adds a complement clause that mentions a reason for not having written the article
(keine ZEIT hatte ‘had no time’, l. 11). This does not clarify the prior reference of
das, but retrospectively shifts it to something only metonymically related to what
was actually said by EG. While the matrix clause signals compliance, the specifica-
tion in the complement clause changes the kind of action that FR commits herself
to. FR then uses the complement clause construction for a turn expansion which
gives a more detailed explication of her prior intention: She contrastively adds a
further complementizer (aber dass ‘but that’, l. 12), but then abandons the struc-
ture. The reformulation in lines 13–15 shows that the reason she gave before is
only one part of the further specification of what she intends to write.
The “janus-faced”, both retrospective and prospective, orientation of the
demonstrative imbues the construction with a rhetorical potential that becomes
clear in cases such as Excerpt (5): The anaphoric pronoun establishes a cohesive
tie with the prior speaker’s perspective; the TCU signals affiliation. This is then
transformed into the announcement of an action which can be more or less at
odds with the prior speaker’s proposal. Still, the shift is not marked in any way, but
the right dislocation makes it appear to be a reformulation.
Excerpt (6) shows how the construction is used for referring to prior actions
of others when there is no verbal antecedent to the pronoun das. A group of men is
playing a “football manager” game, in which they bid on players to build a virtual
team. We join the action when they are bidding on the player Cacau.
(6) “Football manager” game [FOLK_E_00021_T_11, c39]
01 SK zum ERSten,
‘Going once,’
02 (.) zum ZWEIten;
‘going twice,’
03 MT drei SECHS.
‘Three point six [million euros].’
04 DK sagst du auf ENGlisch?
‘Do you say it in English?’
05 (0.23)
06 SK ah:.
‘Ah.’
07 DK drei SIEben.
‘Three point seven.’
08 (1.13)
09 JZ vier.
‘Four.’
10 (.)
11 DK vier FÜNF.
‘Four point five.’
12 (0.82)
13 NI das war doch KLAR dass der nich für_n
3sg.n.nom be.3sg.past modp clear comp 3sg.m.nom not for_det.m.acc
GRUNDpreis über_n lAdentisch geht.

base.price over_det.m.acc counter go.3sg.prs
‘It was obvious that he wouldn’t be sold for the basic price.’
14 (0.43)
15 CH FÜNF millionen.
‘Five million.’
16 (0.94)
17 NI aber sÜß wie der simon_s immer wieder verSUCHT.
‘But it’s sweet how Simon keeps trying.’
SK is trying to end the bidding sequence (l. 1–2), but the others keep going. In
line 13, NI comments on this using the right dislocation construction: The matrix
clause expresses an epistemic stance towards the ongoing interaction and evaluates
it as expectable (das war doch KLAR ‘that/it was obvious’). The complement clause,
which is realized as a continuation of the same intonation contour, summarizes
the bidding activities (nich für_n GRUNDpreis über_n lAdentisch geht ‘not be sold
for the basic/starting price’). At the same time, it specifies the reference of das
in the matrix clause, which otherwise could not be unambiguously interpreted.
By the time das is uttered, it could be interpreted as referring to the immediately
prior turn by DK (vier FÜNF. ‘four point five’, l. 11), or to the whole prior bidding
sequence, or even to something in the non-verbal context. Thus, the interpretation
of the pronoun as not anaphoric but as discourse-deictic25 and at the same time
cataphoric26 is only possible because of the complement clause.
As the examples show, the construction can emerge in the course of construct-
ing a turn online, as in (4) and (5). It can also be used as a whole, as in (6), where
the cohesive prosodic contour and the need to specify the reference of das make
it likely that it was produced (and perceived) as an integrated unit from the start.
In either case it links two potentially separate actions through a cohesive gram-
matical construction. Das thus creates a retrospective link, while at the same time
affording the potential for further (re-)specification. The right dislocation con-
struction thus allows for an incremental constitution of meaning and reference.
Pragmatically, this can be used to reconcile the simultaneous display of affiliation
and implicit disagreement.
25. Both das and the complement clause refer meta-pragmatically to prior actions and not to a
proposition from prior discourse, as in the earlier excerpts.
26. There are 21 cases in the collection that are cataphoric or non-referential. It is not always
easy to determine whether the speaker has intended retrospective, (discourse-)deictic refer-
ence or whether the pronoun is only cataphoric, that is, prospective. These cases show no
special formal characteristics; their prosodic delivery, for example, varies just as much as it does
in the other cases.
4.2 Co-reference with prior parts of the same turn
We now turn to cases in which the demonstrative in the matrix clause refers to an
adjacent or non-adjacent prior TCU in the same turn.27
The first example comes from a public mediation session for the railway con-
struction project “Stuttgart 21”. In Excerpt (7), an expert presents the speed profile
of the planned new trajectory over several TCUs (l. 1–13). He then assesses these
facts positively: des is meine damen und herren ein sehr großer VORteil, (‘This is,
ladies and gentlemen, a huge advantage.’, l. 14). This assessment explicitly turns the
facts, which were presented in a rather neutral fashion, into arguments in favor
of the new trajectory.
(7) Mediation meeting (Stuttgart 21) [FOLK_E_00064_ T_01, c619]
01 VK °h wenn wir uns jetzt mal die schnEllfahrstrecke
Anschauen mit dem geSCHWINdigkeitsprofil;
‘Now, if we take a look at the speed profile of the
high-speed railway line’
02 (0.41) und äh kucken woher KOMMT denn eigentlich der fAhrzeitgewinn?
‘and, um, try to find out where the gain in time comes from,’
03 °h dann muss man SAgen,
‘then one has to say’
04 °h dass wir (0.67) bei der SCHNELLfahrstrecke,
05 (.) hier,
06 (0.43) im hauptbahnhof STUTTgart,
07 (0.36) heRAUSbeschleunigen?
‘that the train on the high-speed line speeds up when
leaving Stuttgart main station,’
08 (0.34) °h dann eine geschwindigkeit von
zweihundertFÜNFzig kilometer pro stunde erREIchen,
‘then gains a speed of 250 km/h.’
09 (.) und diese geSCHWINdigkeit,
‘and this speed,’
10 °h bis auf einen leichten einbruch in dem beREICH,
‘apart from a slight setback in this area,’
11 °h praktisch HALten können,
‘can practically be sustained,’
12 bevor wir dann bei ULM,
13 (.) wieder mit der geschwindigkeit hiNUNtergehen.
‘until we lower the speed again when we reach Ulm.’
14 °h des is meine damen und herren ein
3sg.n.nom be.3sg.prs poss.pl lady.pl and gentleman.pl det.indef.m.nom
27. In 18 of these 35 cases das refers to a non-adjacent prior TCU, in 17 cases to an adjacent one.
sehr großer VORteil,

very big advantage
‘This is, ladies and gentlemen, a huge advantage.’
15 (.) dass wir über die gesamte STRECke,
comp 2pl.nom over det.f.acc whole distance
16 °h eine sehr gleichförmige hohe geSCHWINdigkeit fahren
det.indef.f.acc very constant high speed go.inf
können.
can.2pl.prs
‘that we can go the whole distance with a very constant high speed.’
17 im gegensatz DAzu,
‘In contrast,
18 (.) °h ist auf der HEUtigen strecke,
on today’s route,
19 (0.39) eine geschwindigkeit erreichbar HIER von hundertSECHzig,
a speed of 160 is possible here,‘
20 (.) und sie sehen dass wir überall auf der strecke
laufend und immer wieder EINbrüche haben.
‘and you can see that we have setbacks all over the route.’
The assessment des is meine damen und herren ein sehr großer VORteil, (‘This
is, ladies and gentlemen, a huge advantage.’, l. 14) is syntactically and semanti-
cally complete. The final contour of the intonation unit, however, projects turn
continuation. The added complement clause, dass wir über die gesamte STRECke,
°h eine sehr gleichförmige hohe geSCHWINdigkeit fahren können. (‘that we can go
the whole distance with a very constant high speed.’, l. 15–16), disambiguates the
reference by making clear that it includes all of the speaker’s talk on the issue and
not just a part of it. The complement clause offers a concise summary of it. This
serves as an anchoring point for the following topic expansion, which contrasts the
planned trajectory with the speed restrictions of the current one (l. 17–20). The
right dislocation construction here is a means to coherently relate stance-taking or
assessments to summaries or repetitions of the main argument made before. This
turn-structuring use is found in interaction types with argumentative multi-unit
turns, such as classroom interaction, university vivas, and meetings.
In contrast to such projecting cases, which often seem to be used purposefully,
cases of expansion show how the right dislocation construction emerges out of
the contingencies of online planning and turn-taking. Excerpt (8) comes from a
lesson at a professional school on the topic ‘goals of apprenticeship’. Close to the
end of the lesson, the teacher (GS) wraps up the topic ‘skills’.
(8) Lesson at a professional school [FOLK_E_00004_T_02, c1217]

01 GS im prinzIp ging mir_s HEUte drum,
‘In principle, it was important to me today,’
02 (0.55) ja,
‘right?’
03 (0.37) dass sie (.) einfach (0.22) diese beGRIFfe sich einprägen;
‘that you simply memorize these terms’
04 und damit UMgehe könne;
‘and are able to handle them’
05 (.) und w verSTEhen was man darunter MEINT.=ja?
‘and understand their definitions.’
06 (1.71)
07 GS wenn ma über ausbildungsmethode SPRICHT,
‘If you talk about methods of education’
08 dann ghörn halt diese beGRIFfe noch dazu.
‘then these terms are part of it.’
09 (0.34) ja?
‘Right?’
10 (0.33)
11 GS das is: das zIEl MEIner (.)
3sg.n.nom be.3sg.prs det.n.nom goal det.poss.f.gen
Unterrichtsstunde heut.=ne?
lesson today prtc
‘That’s the goal of my lesson today, right?’
12 (1.07)
13 GS dass sie wissen sie MÜSsen (.)
comp 3pl.nom know.3pl.prs 3pl.nom must.3pl.prs
dem (0.25)azubi kompeTENze vermittle;
det.M.dat apprentice.dat skills.acc teach.inf
‘that you know you have to teach an apprentice certain skills’
14 (0.37)
15 GS un was damit geMEINT is.
and 3sg.int there_with mean.ptcp aux.pass.3sg.prs
‘and what is meant by that.’
16 (2.25)
17 GS vor dem begriff kompetEnze hatte ma scho_ma_n beGRIFF,
‘Before the term ‘skills’ we had talked about another term,
18 (0.33) ne?
right?’
19 (0.63)
20 GS dis ÄNdert sich so ab und zu ma,
‘- That changes from time to time. –’
21 (2.84) der nannte sich SCHLÜSselqualifikatione.
‘That one was called key qualifications.’
GS explains that the aim of the last exercise was to familiarize the participants
with certain technical terms (l. 1–9). She refers back to what she has said before
with das and classifies it as das zIEl MEIner (.) Unterrichtsstunde heut. (‘the goal of
my lesson today.’, l. 11), creating a TRP, which is emphasized by the question tag
ne? (‘right?’, l. 11). After a pause (l. 12), GS uses an expansion with a complement
clause (l. 13–15) to re-phrase her goals. She offers a concrete example (kompe-
TENze ‘skills’, l. 13) for the aforementioned abstract term beGRIFfe (‘terms’ or
‘concepts’, l. 3 and 8).28 This is used as a link (l. 17) to a next closely related topic
SCHLÜSselqualifikatione ‘key qualifications’, l. 21).
Just as when responding to a prior turn, in turn-internal use the right disloca-
tion construction serves as a conventionalized format for tying a backward-ori-
ented action and a specification or shift of reference together in one grammatical
construction—either pre-planned by the speaker or emergently. In contrast to the
uses referring to (parts of) another speaker’s prior turn, the cases referring to parts
of a speaker’s own turn less often involve a shift to more remote referents. Rather,
the construction is used for repetitions or specifications with only a slight shift in
reference (as in (8)).
4.3 Co-constructed uses: Referential self-repair and understanding-check
Co-constructed uses of the right dislocation construction are rare in our data. Yet,
they give particularly strong evidence for its conventionalized status. We will look
at one example in which both parts of the construction are produced by the same
speaker and the complement clause is a referential self-repair triggered by another
speaker’s repair initiation, and another example in which the matrix clause is
produced by one speaker and the complement clause by another speaker as an
understanding-check.
Excerpt (9) comes from a meeting at a social institution. A group of social
workers discuss the activities of a boy who is cared for at their institution. SZ ar-
gues that it is important that the boy keeps attending his hapkido lessons, because
the trainer accepts him as he is, integrates him well, and does not put pressure on
him, as she suspects might be the case at other sports clubs (l. 1–13).
(9) Meeting at a social institution [FOLK_E_00022_T_03, c1172]
01 SZ und deswegen find ich_s wIchtig ähm dass er dA echt daBEIbleibt;
‘And that’s why I think it’s important that he keeps going there.’
28. Kompetenzen (‘skills’) have been mentioned earlier, in a previous lesson at the professional
school. It is itself a cover term for several specific skills (e.g., Methodenkompetenz ‘method skill’)
that were discussed in an earlier exercise in the current lesson.
02 (.) weil dis halt auch noch ne grUppe isch wo geNAU,

‘Because this is a group where
03 (0.82) wo der finn aber (0.24) mit gUtem gewissen (.)
so wie er Isch hingehen KANN.
where Finn can go as he is without concern.’
04 (0.72)
05 SZ weil da net wie in nem üblichen SPORTverein oder so so_n ähm–
‘Because unlike in a normal sports club or something there’s no
06 (0.22)
07 AW LEIStungsdruck.
‘Pressure.’
08 SZ mObbing oder leischtungsdrutsch äh druck STATTfindet.
‘mobbing or pressure there.’
09 (0.95)
10 SZ und w[eil] äh weil_s auch_n trainer isch der die die
kinder gAnz andersch irgendwie äh im BLICK hat;
‘And because there’s a trainer who monitors the children well
11 HM [ja.]
‘Yeah.’
12 SZ und Ihn auch (0.57) auf sEIne art also so mit EINbezieht un so;
and who integrates him in his own way.’
13 un dis gibt_s woandersch (0.35) [halt eher] (.) oder_s
gibt_s halt SELten.
‘And that’s a rare thing elsewhere.’
14 HM [hm_m. ]
15 SZ °h
16 HM [hm:. ]
17 AW [hast du] ihm dis bei
aux.2sg.prs 2sg.nom 3sg.M.dat 3sg.n.acc at
eurem gespräch so geSAGT?
poss.2pl.dat conversation this_way say.ptcp
‘Did you tell him that in your conversation?’
18 (0.2)
19 SZ WAS denn?
‘[Tell him] what?’
20 (1.6)
21 AW ähm,
22 (0.28) dass er da sO ANgenommen
comp 3sg.m.nom there as take.ptcp
wird wie er IS?
aux.pass.3sg.prs as 3sg.m.nom be.3sg.prs
‘Um, that he is accepted the way he is there.’
23 (0.73)
24 AW °h äh [ich hab jetzt ]
‘Um, now I’
25 SZ [nee weil er hat] ja zu mIr gar nix geSA[GT ] davon.

‘No, because he never said anything about it to me.’
26 AW [ah,]
27 alles KLAR.
‘I see.’
The reference of the demonstrative in AW’s question hast du ihm dis bei eurem
gespräch so geSAGT? (‘Did you tell him that in your conversation?’, l. 17) can be
interpreted as wide, encompassing SZ’s whole prior turn. But it can also be inter-
preted as referring to only a part of it, for example the immediately adjacent TCU
in line 13. Accordingly, SZ initiates a referential repair by asking WAS denn? (‘[Tell
him] what?’, l. 19). For her repair, AW uses the structural latency (Auer, 2015) of
the syntax of her own prior turn and expands it with a complement clause. Thus,
the non-projected expansion disambiguates the reference of the demonstrative. It
is a formulation of SZ’s TCU in line 3, but it also gives a concise summary of SZ’s
whole turn in lines 2–12; yet, it excludes the possibility that the question refers
to the TCU in line 13.
In contrast to this other-initiated expansion, in Excerpt (10) a second speaker
expands the first speaker’s turn with a complement clause to check whether she
has correctly recovered what the first speaker referred to. The example comes from
the same telephone conversation as extracts (1) and (5).
(10) Telephone conversation among friends [FOLK_E_00084_T_02, c120]
01 EG sind eigentlich irgendwie (0.21) grad viele themen so
OFfen immer wieder,
‘A lot of topics are “open” at the moment, actually.’
02 °h h°
03 (0.28)
04 FR ja eli[f hat doch voll] VIEle–
‘Yeah, Elif is working on a lot of them
05 EG [hm_hm. ]
06 FR IMmer.
always.’
07 [is mir auch schon AU]fgefallen.
‘I’ve also noticed that.’
08 EG [hm_m. ]
09 FR °h die hat ja auch immer die sagt auch immer so SAchen bei ähm–
‘And she always says things during
10 (0.7) den normalen TAgessitzungen–
the normal daily_meetings.’
11 irgendwie themen die OFfen sind und;
‘Topics that are “open”. And’
12 °h die hängt sich da voll REIN. h°
she puts a lot into it.’
13 (0.84)
14 FR °h aber die meisten finden das NICH so geil,
but det.nom.pl most find.3pl.prs 3sg.n.acc not so great
15 hab ich den EINdruck?
have.1sg.prs 1sg.nom det.acc.sg impression.acc
‘But most [of our colleagues] do not find that so
cool, I have the impression.’
16 (1.22)
17 EG dass wir jetz diese eigenen THEmen machen?
comp 2pl.nom now det.dem.pl.acc own.pl.acc topic.pl make.2pl.prs
‘That we [are asked to] work on our own topics now?’
18 (0.23)
19 FR ja?
‘Yes.’
20 EG °h ja.
‘Yeah.’
21 (0.78) viele ham da halt kein BOCK drauf.
‘Many [of them] don’t want to do it.’
EG and FR are talking about the editorial meetings of the newspaper for which
they work. EG mentions that there are many ‘open’ topics on the agenda (i.e., top-
ics that the writers have chosen themselves) and that one colleague, Elif, is putting
a lot of energy into working on many of them (l. 1–12). FR states aber die meisten
finden das NICH so geil, hab ich den EINdruck? (‘But most [of our colleagues] don’t
find that so cool, I have the impression.’, l. 14–15). The rising intonation indicates
that she expects EG to affiliate with her. A pause follows, which can be taken to
indicate that it is not clear whether FR is referring to the existence of ‘open’ topics
(l. 1) or to Elif’s efforts to deal with them (l. 4 and 12). EG then produces the
complement clause dass wir jetz diese eigenen THEmen machen? (‘That we [are
asked to] work on our own topics now?’, l. 17) as a check of understanding.29 FR
confirms EG’s interpretation with ja? (‘yes’, l. 19) with a rising contour. This again
pursues EG’s response, which she finally delivers (l. 20–21).
In the example, the complement clause is not a collaborative completion (as
in Günthner’s [2015: 44] example for a ‘free-standing’ dass-clause that fulfils the
projection of another speaker’s matrix clause).30 Here, the first speaker’s clause
only retrospectively becomes a matrix clause, because it has been syntactically,
prosodically, semantically, and pragmatically complete. The complement clause is
29. For understanding-checks with es in the matrix clause see Günthner (2015: 52ff.), and with
das in the matrix clause see Günthner (2013: 232ff., 2015: 55).
30. The expletive pronoun in the matrix clause in Günthner’s example is es, not das. Comparable
cases with the demonstrative seem possible introspectively, although our collection does not
contain any.
not projected and performs an action of its own. This is also true for the comple-
ment clause used as a self-repair (produced as an answer to a question) in (9).
In contrast, in the cases that are not co-constructed, it is disputable whether the
complement clause performs a separate action, that is, if ‘specification’ counts as
such (see Section 5).
5. Discussion
Our analyses have shown that the right dislocation construction is a conventional
pattern of spoken German: Prosodically, it can occur both as a coherently pro-
duced whole and it can be produced in two chunks, that is, separate prosodic
units which index the potentially independent status of the first clause and the two
actions that they accomplish. The interactively constructed cases make especially
clear that each clause can accomplish an action of its own (the complement clause
can, for example, be a repair initiation or a repair as in (9) and (10)). In the cases
that are not co-constructed, the matrix clause performs a complete action inde-
pendently before the complement clause is added. It remains debatable whether
the complement clause merely continues that action or whether ‘(re-)specification’
is an independent action. What speaks in favor of the independent action status
is that specifications of potentially ambiguous pronominal references frequently
occur as independent main clauses as well (see e.g., Excerpt (5), l. 5–631). The
right-dislocation construction thus is one grammaticalized way to produce the
two actions of asserting (or, more specifically, for example, assessing or asking
a question) and specifying reference as successive chunks of a multi-unit turn
(see Couper-Kuhlen & Thompson, 2008 for a comparable view on the English
extraposition construction).
The right dislocation pattern is very productive—a great variety of verbs are
used in this construction. In contrast to what Günthner (2009) has found for
extraposition, there is no tendency towards the formulaicity of the matrix clause.
While many of them have epistemic, evidential or evaluative semantics, event-
describing verbs are used as well (e.g., das erledigt das LAbel; dass die uns bei der
GEma registrieren. ‘The label handles this for us, that they register us with GEMA.’;
FOLK_E_00044_ T_01, c1007). Yet the tendency to contain verbs with ‘subjective’
semantics is common to all matrix clause constructions (see Section 2); it simply
arises from the fact that many complement-taking predicates have meanings
expressing attitudes towards propositional referents.
31. Lines 5–6 of Excerpt (5): dann MACHT er das auch nich; °h dann GIBT er das auch keinem
andern. ‘Then he won’t do that. Then he won’t give it [your topic] to anyone else.’
The actions performed by the matrix clause are accordingly diverse as well, as
the examples have shown. Questions and assessments make up important parts
of the collection, but are on the whole not more frequent than other actions.32
Thus, as has been found for right-dislocated NPs in French by Horlacher (2015)
and Pekarek Doehler et al. (2015), matrix clauses of right-dislocated complement
clauses in German are frequently used for, but not specifically tied to, assess-
ments.33 Rather, the bi-clausal right dislocation pattern is a very flexible construc-
tion, whose overarching characteristics (‘chunking’ and ‘linking’, incremental (re-)
specification of meaning, and potentially retrospective and prospective reference)
can be used in the service of various actions.
6. Conclusion
The combination of matrix clause and right-dislocated complement clause is a

linking construction that has retrospective and prospective properties. The first
clause performs a backward-oriented action and the second clause (re-)specifies
reference, but also often includes lexical material that prospectively serves as a link
to the content of further, upcoming TCUs. The construction is thus particularly
suited for linking parts of multi-unit turns.
The construction is conventionalized and thus can be used as a whole. Yet it
often results from online planning and expansions, taking advantage of structural
latencies of the emerging turn structure (Auer, 2015), in particular the potential to
reinterpret the (originally anaphoric) pronoun. Because of its bi-partite structure,
the construction allows for the incremental production of meaning, that is, a suc-
cessive realization of an action (for example, a statement, assessment, or question)
and a specification/disambiguation of complex propositional or meta-pragmatic
references, which can yield a summary, a change of perspective, or a topic expan-
sion. The construction creates cohesion and accommodates emergent interactional
contingencies of structuring multi-unit turns, securing understanding or (more or
less strategic) perspective shifts. From the point of view of information structure,
it adheres to Chafe’s (1994) ‘one new idea’ constraint, presenting only one new con-
cept or referent per clause and often also per intonation unit. Thus, it is part of a
large family of bi-clausal constructions (such as extrapositions and pseudo-clefts;
see e.g., Günthner & Hopper, 2010) that have been described in the literature as
produced in separate chunks that perform separate functions and accommodate
the temporality of talk-in-interaction.
32. Among the 93 cases in the collection are 21 assessments and 10 questions.
33. The other sequential functions described as typical in the literature on French are not
frequent in our collection.
References
Ahrenholz, Bernt. 2007. Verweise mit Demonstrativa im gesprochenen Deutsch. Grammatik,

Zweitspracherwerb und Deutsch als Fremdsprache. Berlin: de Gruyter.
https://doi.org/10.1515/9783110894127
Altmann, Hans. 1981. Formen der „Herausstellung“ im Deutschen: Rechtsversetzung, Linksverset-
zung, freies Thema und verwandte Konstruktionen. Tübingen: Niemeyer.
https://doi.org/10.1515/9783111635286
Auer, Peter. 1991. Vom Ende deutscher Sätze. Zeitschrift für germanistische Linguistik 19,
139–157. https://doi.org/10.1515/zfgl.1991.19.2.141
Auer, Peter. 1996. On the prosody and syntax of turn-continuations. In Elizabeth Couper-Kuhlen
& Margret Selting (Eds.), Prosody in conversation. Interactional studies. (pp. 57–100). Cam-
bridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511597862.004
Auer, Peter. 2005. Projection in interaction and projection in grammar. Text 25(1), 7–36.
Auer, Peter. 2007. Why are increments such elusive objects? An afterthought. Pragmatics 17(4),
647–658. https://doi.org/10.1075/prag.17.4.03aue
Auer, Peter. 2010. Zum Segmentierungsproblem in der Gesprochenen Sprache. InLiSt – Interac-
tion and Linguistic Structures, 49. http://www.inlist.uni-bayreuth.de/issues/49/InList49.pdf
Auer, Peter. 2015. The temporality of language in interaction: Projection and latency. In Arnulf
Deppermann & Susanne Günthner (Eds.), Temporality in interaction (pp. 27–56). Amster-
dam: Benjamins.
Averintseva-Klisch, Maria. 2009. Die NP-Rechtsversetzung im Deutschen. Tübingen: Stauffen-
burg.
Barth-Weingarten, Dagmar. 2011. Double sayings of German JA – more observations on their
phonetic form and alignment function. Research on Language and Social Interaction 44(2),
157–185. https://doi.org/10.1080/08351813.2011.567099
Chafe, Wallace L. 1976. Givenness, contrastiveness, definiteness, subjects and topics. In Charles
N. Li (Ed.), Subject and topic (pp. 27–55). New York, NY: Academic Press.
Chafe, Wallace. 1994. Discourse, consciousness, and time: The flow and displacement of conscious
experience in speaking and writing. Chicago, IL: The University of Chicago Press.
Consten, Manfred, & Schwarz-Friesel, Monika. 2007. Anapher. In Ludger Hoffmann (Ed.),
Handbuch der deutschen Wortarten (pp. 265–292). Berlin: de Gruyter.
Consten, Manfred, Knees, Mareile, & Schwarz-Friesel, Monika. 2007. The function of complex
anaphors in texts: Evidence from corpus studies and ontological considerations. In Monika
Schwarz-Friesel, Manfred Consten, & Mareile Knees (Eds.), Anaphors in text: Cognitive,
formal and applied approaches to anaphoric reference (pp. 81–102). Amsterdam: Benjamins.
https://doi.org/10.1075/slcs.86.09con
Couper-Kuhlen, Elizabeth, & Ono, Tsuyoshi (Eds.). 2007. Turn-continuation in cross-linguistic
perspective: Introductory remarks. Special Issue of Pragmatics 17(4).
Couper-Kuhlen, Elizabeth, & Thompson, Sandra A. 2006. You know, it’s funny: Eine Neube-
trachtung der Extraposition im Englischen. In Susanne Günthner & Wolfgang Imo (Eds.),
Konstruktionen in der Interaktion (pp. 23–58). Berlin: de Gruyter.
https://doi.org/10.1515/9783110894158.23
Couper-Kuhlen, Elizabeth, & Thompson, Sandra A. 2008. On assessing situations and events in
conversation: ‘Extraposition’ and its relatives. Discourse Studies 10(4), 443–467.
https://doi.org/10.1177/1461445608091882
Deppermann, Arnulf, & Proske, Nadine. 2015. Grundeinheiten der Sprache und des Sprechens.
In Christa Dürscheid & Jan-Georg Schneider (Eds.), Handbuch Satz, Äußerung, Schema
(pp. 17–47). Berlin: de Gruyter. https://doi.org/10.1515/9783110296037-003
Du Bois, John W. et al. 1993. Outline of Discourse Transcription. In Jane A. Edwards & Martin D.
Lampert (Eds.), Talking Data: Transcription and Coding in Discourse Research (pp. 45–89).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Ford, Cecilia, & Thompson, Sandra A. 1996. Interactional units in conversation: Syntactic, into-
national, and pragmatic resources for the management of turns. In Elinor Ochs, Emanuel
A. Schegloff, & Sandra A. Thompson (Eds.), Interaction and Grammar (pp. 134–184). Cam-
Gallmann, Peter. 2009. Der Satz. In Matthias Wermke, Kathrin Kunkel-Razum, & Werner
Scholze-Stubenrecht (Eds.), Duden – die Grammatik. 8., überarbeitet Auflage. Mannheim:
Dudenverlag.
Geluykens, Ronald. 1987. Tails (right dislocations) as a repair mechanism in English conversa-
tion. In Jan Nuyts & Georges De Schutter (Eds.), Getting one’s word into line: On word order
and functional grammar (pp. 119–129). Dordrecht: Foris.
Golato, Andrea, & Fagyal, Zsuzsanna. 2008. Comparing single and double sayings of the Ger-
man response token ja and the role of prosody: A Conversation Analytic perspective.
Research on Language and Social Interaction 41(3), 241–270.
https://doi.org/10.1080/08351810802237834
Günthner, Susanne. 2009. Extrapositionen mit es im gesprochenen Deutsch. Zeitschrift für
germanistische Linguistik 37, 15–47. https://doi.org/10.1515/ZGL.2009.003
Günthner, Susanne. 2013. Vom schriftsprachlichen Standard zur pragmatischen Vielfalt? As-
pekte einer interaktional fundierten Grammatikbeschreibung am Beispiel von dass-Kon-
struktionen. In Jörg Hagemann, Wolf Peter Klein, & Sven Staffeldt (Eds.), Pragmatischer
Standard (pp. 223–243). Tübingen: Stauffenburg.
Günthner, Susanne. 2015. Geteilte Syntax: Kollaborativ erzeugte dass-Konstruktionen. In Al-
exander Ziem & Alexander Lasch (Eds.), Konstruktionsgrammatik IV. Konstruktionen als
soziale Konventionen und kognitive Routinen (pp. 25–40). Tübingen: Stauffenburg.
Günthner, Susanne, & Hopper, Paul J. 2010. Zeitlichkeit & sprachliche Strukturen: Pseudoclefts
im Englischen und Deutschen. Gesprächsforschung – Online-Zeitschrift zur verbalen Inter-
aktion 11, 1–28. http://www.gespraechsforschung-online.de/fileadmin/dateien/heft2010/
ga-guenthner.pdf
Horlacher, Anne-Sylvie. 2015. La dislocation à droite revisitée: une approche interactionniste.
Louvain-la-Neuve: de boeck. https://doi.org/10.3917/dbu.hcher.2015.01
Keevallik, Leelo. 2011. Pro-forms as projective devices in interaction. Discourse Processes 48,
404–431. https://doi.org/10.1080/0163853X.2011.559150
Ono, Tsuyoshi, & Thompson, Sandra A. (Eds.). 2012. A cross-linguistic, multi-modal, gram-
maticization perspective on turn continuation in conversation. Special issue of Discourse
Processes 49(3–4).
Pekarek Doehler, Simona, De Stefani, Elwys, & Horlacher, Anne-Sylvie. 2015. Time and emer-
gence in grammar. Dislocation, topicalization and hanging topic in French talk-in-interaction.
Amsterdam: Benjamins. https://doi.org/10.1075/slsi.28
Proske, Nadine. 2013. Informationsmanagement im gesprochenen Deutsch. Eine diskurspragma-

tische Untersuchung syntaktischer Strukturen in Alltagsgesprächen. Heidelberg: Winter.
Ross, John Robert. 1967. Constraints on variables in syntax. Ph.D. dissertation, MIT.
Schegloff, Emanuel A. 1996. Turn organization: One intersection of grammar and interaction.
In Elinor Ochs, Emanuel A. Schegloff, & Sandra A. Thompson (Eds.), Interaction and
Grammar (pp. 52–133). Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511620874.002
Schmidt, Thomas. 2014. The research and teaching corpus of spoken German – FOLK. In
Proceedings of the Ninth conference on International Language Resources and Evaluation
(LREC’14). Reykjavik, Iceland: European Language Resources Association (ELRA).
Selting, Margret et al. 2009. Gesprächsanalytisches Transkriptionssystem 2 (GAT 2). Ge-
sprächsforschung – Online-Zeitschrift zur verbalen Interaktion 10, 353–402. http://www.
gespraechsforschung-ozs.de/heft2009/px-gat2.pdf
Uhmann, Susanne. 1993. Das Mittelfeld im Gespräch. In Marga Reis (Ed.), Wortstellung und
Informationsstruktur (pp. 313–354). Tübingen: Niemeyer.
https://doi.org/10.1515/9783111658469.313
Zifonun, Gisela, Hoffman, Ludger, & Strecker, Bruno (Eds.). 1997. Grammatik der deutschen
Sprache. Berlin: de Gruyter.
Chapter 11
Relative-clause increments and

the management of reference
A multimodal analysis of French talk-in-interaction
Ioana-Maria Stoenica and Simona Pekarek Doehler
In this paper we propose a reanalysis of relative clauses in French talk-in-

interaction as part of “grammar for talk implementing action” (Schegloff,
1996: p. 113). Our analytic focus is on relative clauses produced as increments,
i.e., cases where the [main clause + relative clause] pattern emerges gradually,
in response to interactional contingencies such as co-participants’ verbal and
embodied conduct. We identify two recurrent interactional purposes that speak-
ers accomplish by means of such self-incremented relative clauses: referential
repair, ensuing from a recipient’s verbal and/or embodied display of trouble;
referential elaboration, ensuing from a recipient’s verbal and/or embodied display
of referent recognition. The findings challenge the notion of relative clauses as
subordinate clauses, and extend our understanding of the emergent nature of
grammar to the field of complex syntax.
Keywords: relative clauses, increments, grammar-in-interaction, embodied

conduct, repair
1. Introduction1
The locally adaptive and emergent nature of grammar has been empirically at-
tested, throughout the past two or three decades, by several lines of research on
discourse and social interaction. This research has shown that the constituency
of grammatical constructions and of larger syntactic trajectories is configured in
real time, moment-by-moment (Goodwin, 1979; Hopper, 1987, 2011; Auer, 2009;
1. We thank Leelo Keevallik, Yael Maschler and Ritva Laury for their insightful comments on
an earlier version of this paper. The present study was carried out with the generous support of
the Swiss National Science Foundation, grant no. 100012_178819.
https://doi.org/10.1075/slsi.32.11sto
304 Ioana-Maria Stoenica and Simona Pekarek Doehler
Deppermann and Günthner, 2015; Pekarek Doehler et al., 2015), and that such lo-
cal emergence may lead not only to on-line expansions but also to on-line revisions
of syntactic trajectories, involving change in constituent status (Pekarek Doehler,
2011; Norén, 2013). When it comes to studies on complex syntax, however, the way
clause-combining patterns emerge in real time has gained comparatively little atten-
tion (but see e.g., Couper-Kuhlen, 2012). Most studies have so far focused on how
pieces of canonical clause-combining patterns occur as syntactically independent
stretches of talk (see e.g., Günthner, 2008; Hopper and Thompson, 2008 and some
of the papers in Laury and Suzuki, 2011 on projector constructions, and Thompson,
2002; Kärkkäinen, 2003 and the papers in Lindström et al., 2016 on complement-
taking-predicate constructions), or on how complementizers, rather than linking
clauses, function as particles (see the papers in Laury, 2008). As a consequence,
we still know little about how ‘simple syntax’ expands into ‘complex syntax’ in real
time, that is, how clause-combining patterns emerge incrementally (cf. Schegloff,
1996; Ford et al., 2002), out of the moment-by-moment unfolding of turns and
actions, as part of people’s local meaning-making processes in social interaction.
In this paper we document how complex syntactic patterns in use may result
from local contingencies of social interaction, in response to recipients’ actions
or absence of these. We propose a reanalysis of relative clauses (RCs) as part of
“grammar for talk implementing action” (Schegloff, 1996: p. 113), based on close
scrutiny of the sequential deployment of participants’ verbal and embodied
conduct. We focus on RCs produced as self-increments, that is, on cases where
a speaker extends his or her own turn past a transition relevance place (TRP) by
means of a relative clause. We identify the kind of interactional contingencies that
the self-incremented RCs respond to and the precise purposes they are used to
fulfill. While we stress that the prosodic delivery of the clause-combining pattern
is decisive for understanding its locally emergent nature (cf. Couper-Kuhlen,
1996, 2012), we also argue that close scrutiny of co-occurring embodied conduct
is indispensable for identifying its interactional workings. In doing so, we work
toward a more holistic understanding of grammar as it occurs, emerges, and func-
tions within the multimodal ecology of social interaction.
This paper is part of a growing body of research that empirically documents
how detailed analysis of embodied conduct can enrich our understanding of
grammar-in-interaction (Goodwin, 1979; Keevallik, 2018). It is designed to provide
novel insights into both the interactional workings of RCs and the functioning of
increments. By focusing on a specific grammatical form of increments, the paper
responds to the currently increasing interest in the precise grammatical make-up
of increments, and how this make-up accounts for the distinctive interactional
workings of increments (cf. Clift, 2007; Seppänen and Laury, 2007; Horlacher,
2015). Based on a praxeological approach to RCs as resources for (inter-)action,
Chapter 11. Relative-clause increments and the management of reference 305
the paper challenges the notion of RCs as subordinate clauses in the light of the
syntax for interaction, and documents how RCs are used in locally contingent
ways to accomplish actions in talk-in-interaction. Last but not least, by scrutiniz-
ing local contingencies of the use of RCs, the paper seeks to contribute to a better
understanding of the emergent nature of clause-combining in interaction, adding
to the rich body of evidence that supports an understanding of grammar as an
emergent (Hopper, 1987, 2011) and adaptive resource for action.
In what follows, we first detail the object of our study (Section 2) and present
data and procedures (Section 3). We then document two recurrent interactional
purposes that self-incremented RCs are used to accomplish in French talk-in-inter-
action, namely referential repair (Section 4) and referential elaboration (Section 5).
By referential elaboration, we understand the work deployed by a speaker to provide
further information about a prior referent, but not designed as doing repair, that
is, not responding to trouble. We show that the accomplishment of these actions
responds to two distinct interactional contingencies: A co-participant’s display
of trouble with referent recognition on the one hand; a co-participant’s display
of referent recognition on the other hand – both displays materializing through
verbal or non-verbal conduct, or a combination of both. Based on the analyses,
we argue that clause-combining routinely appears as an emergent product of how
people deal with practical problems in interaction (Section 6).
2. Relative clauses and increments
2.1 Relative clauses
The clause-combining pattern known as [main clause + relative clause] is amongst

the most classically studied exemplars of complex syntax in the history of lin-
guistics. RCs themselves count as prototypical examples of subordination (but see
below), and they have been abundantly investigated, from a range of theoretical
perspectives, as to their semantic as well as syntactic properties (e.g., Givón,
1990, 1995 for English; Gapany, 2004 for French). Up to now, however, the social-
interactional workings of RCs have remained largely unexplored (but see Tao and
McCarthy, 2001 and Clift, 2007 on English, Laury and Helasvuo, 2015 on Finnish),
and so has their relation to other semiotic resources such as gesture, posture, and
gaze (but see Stoenica, 2018 on French).
RCs have been traditionally defined as subordinate clauses introduced by
a relative pronoun. They are understood to maintain, via the relative pronoun,
a referential relationship with a noun phrase (NP) – the so-called ‘head of the
RC’ – occurring in the main clause (for French, see Riegel et al., 2016). In French,
the relative pronoun is compulsory, as opposed to English (see Fox and Thomson,
2007), where it is compulsory only when it is the subject of the RC. Also, in
French, the relative pronoun is marked for case: qui is used for subjects, que for
direct objects, as illustrated in Excerpts (1) and (2) respectively (RCs are marked
in bold); there are other forms, such as à qui/quoi, auquel, de qui/quoi, duquel and
dont for indirect (prepositional) objects, and où for ‘where’:
(1) Codi_SecII_L2_EO-1_27:00
01 LUC: et puis elle tente d- d’éliminer un peu
‘and then she tries t- to eliminate a bit the’
02 la vérité (0.3) qui est représentée par

the truth pro.rel-sbj be.3sg represented by
‘the truth which is represented by’
03 la personne d’Inès.
det person of Inès
‘the person of Inès’
(2) Codi_SecII_L2_EO-1_06:40
01 ENS: on s’intéresse pas du tout aux fautes
‘we are not at all interested in the errors’
02 que les personnes font en fait.

pro.rel-do det persons make in fact
‘that people make in fact’
The [main clause + relative clause] pattern has been the object of extensive re-
search. One important result of earlier research on complex sentences in general
is the deconstruction of a binary understanding of subordination vs. coordina-
tion in favor of a non-discrete view of clause linkage (Foley and Van Valin, 1984;
Matthiesen and Thompson, 1988; Evans, 2007), which has also been discussed in
research on naturally occurring spoken interaction (Günthner, 1996; Auer, 1998).
This resonates with some more recent studies on spoken French (Berrendonner
and Béguelin, 1989; Deulofeu, 1999; Berrendonner, 2008) that have called into
question the subordinate status of certain RCs, describing these instead as
independent clauses that are pragmatically related to the preceding clause. For
instance, Béguelin et al. (2000) quote the following example:
(3) Béguelin et al. 2000: 307
IBM Suisse a pris l’affaire très au sérieux, qui a prévenu par lettre tous ses
clients.
‘Swiss IBM took the matter very seriously, who informed by letter all its clients.’
Here, the RC is separated from its head NP (Swiss IBM) and placed at the end of
the clause containing that head NP.2Béguelin et al. (2000) argue that in such cases
RCs are very close to coordination. Along similar lines, Jespersen (1933) suggested
earlier on that in such cases the relative pronoun corresponds to ‘and + personal
pronoun’ (e.g., ‘who’ = ‘and he/she/they’). This implies that the relative pronoun
merely indexes an anaphoric link (Brunner, 1981). The Groupe de Fribourg (2012)
documents clauses relatives autonomes, that is, syntactically autonomous relative
clauses, amongst which they count cases where two RCs, despite the fact that they
are coordinated by but or by and, relate to two different head NPs. This is illus-
trated in the following example, in which the first RC (‘that the gang has admitted
to’) relates to ‘eleven hold-ups’, while the second RC (‘who must have committed
many others’), preceded by but, relates to a different NP, namely ‘the gang’.
(4) Groupe de Fribourg 2012: 264
Onze hold-up que la bande a reconnus, mais qui a dû en commettre bien
d’autres.
‘Eleven hold-ups that the gang has admitted to, but who must have committed
many others.’
While taking stock of the above developments, we depart from existing work by
proposing an interactional approach to the study of the [main clause + relative
clause] pattern. Studies on the working of RCs in social interaction are still rare.
Existing work has been interested in the form, the interactional function, and
the sequential placement of RCs (see e.g., Fox and Thompson, 2007 and Tao and
McCarthy, 2001 on English; Laury and Helasvuo, 2015 on Finnish; Maschler, 2011
on Hebrew). Clift (2007) studied the use of English RCs as increments, i.e., as
syntactic continuations of turns that have already reached a syntactic, pragmatic
and prosodic completion (cf. Schegloff, 1996; Ford and Thompson, 1996); she
shows that, by means of RC increments, speakers accomplish assessments and
pursue alignment on the part of recipients. Stoenica (2014) analyzed the use of
‘recycled’ RCs in French conversations, showing that speakers can produce second
or third RCs (thereby recycling their own or a preceding speaker’s RC) in pursuit
of recipient uptake or for displaying affiliation (see also Stoenica 2016a and b).
Stoenica (2018) also documented a range of interactional workings of RCs, in-
cluding stance-taking, initiating referential repair, or adding supplementary items
to enumerations of referent-related characteristics. To our knowledge, Stoenica’s
work is the first set of studies systematically scrutinizing the interactional work-
ings of RCs in relation to participants’ embodied conduct.
2. Givón (1990) referred to these as “extraposed relative clauses”.

2.2 Relative clauses as increments
In this paper, we pursue a similar line of investigation by studying RCs within

their naturally occurring multimodal ecology of use. We show that RCs present a
practical resource for incrementation, as illustrated in Excerpt (5):
(5) ‘Que Romain il nous a parlé’ [Corpus Pauscaf (Pause 18): 02m10-02m18]
01 DAV: j’ai regardé le film Safe.
‘I’ve watched the movie Safe’
02 (0.6)
03 DAV: que Romain il nous a parlé une fois.=

pro.rel-do Romain he us aux tell.pp det time
‘that Romain once told us about’
David’s turn at line 01 ends on a complex TRP (Ford and Thompson, 1996).
Following a 0.6s pause, the speaker then extends his turn (l. 03) by adding syn-
tactically linked material – an increment (Schegloff, 1996, 2000, 2016; Walker,
2001; Ford et al., 2002; Lindström, 2006; Couper-Kuhlen and Ono, 2007) – in the
form of an RC (note that the form of the relative pronoun does not conform to
French standard grammar, which would require dont ‘about which’ rather than
que ‘that/which’). Retrospectively, lines 01 and 03 taken together can be heard as a
single sentential pattern of the [main clause + relative clause] type, that reads ‘I’ve
watched the movie Safe that Romain once told us about’. While Excerpt (5) shows
a so-called ‘post-gap increment’ (Schegloff 2000, 2016), RC increments also occur
in the data as post-other talk increments (see Section 4 below).
According to Schegloff (1996: p. 90), increments are:
elements of talk added to the TCU and the turn which re-occasion possible
completion; that is, which constitute extensions to the TCU or the turn (the
two are different) and which themselves come to another possible completion
of the TCU or turn.
Rather than doing a new action, increments are understood by Schegloff as pro-
longing the action of the extended turn, so as to, for instance, create an additional
sequential opportunity for recipient reaction. Such an interpretation seems at first
sight to hold for Excerpt (5) quoted above, where the RC in line 03 may be heard
as a re-completion (Tanaka, 1999), providing a second sequential opportunity for
recipient response, after a lack of such response (l. 02), to the informing provided
in line 01. However, as we will show in Section 4, the examination of the partici-
pant’s embodied conduct in Excerpt (5) sheds a different light on the interactional
working of the RC increment, suggesting that the incrementally added RC accom-
plishes an action in itself, rather than re-doing or extending a preceding action.
Importantly, the distinction between extending a TCU and starting a new

TCU is sometimes tricky to draw (e.g., Couper-Kuhlen, 2012). Also, as Auer ar-
gues, “it is not clear to what extent increments can be said to constitute actions of
their own” (Auer, 2007: p. 650). We document in our analysis that RC increments
represent TCUs and actions in their own right. We also show that RC increments
accomplish a range of specific interactional jobs that have not been documented
for other types of increments, hence shedding further light on how the gram-
matical material matters for what increments do in social interaction (cf. Pekarek
Doehler and Horlacher, 2013; Horlacher and Pekarek Doehler, 2014; Horlacher,
2015; Stoenica, 2018; Stoenica et al. to appear). And we provide evidence for
how increments respond to local interactional contingencies that materialize not
through verbal but through embodied conduct, such as a recipient’s frowning
and/or gazing into the void.
3. Data
The data for this study consist of 27 video-recorded ordinary conversations in a

university cafeteria (total = 9h), each comprising 2 to 3 participants (all but one
are students). Participants were recorded from two different angles while sitting at
a table, drinking coffee or tea, and/or eating a snack. The data are coincidentally
balanced in terms of gender (33 male, 31 female participants). They were comple-
mented by 10h of audio-recorded informal focus-group interactions, comprising
4 to 5 participants. We drew selectively on the audio data in those cases where the
video data did not yield robust enough collections. All data were transcribed ac-
cording to CA conventions; transcripts of selected excerpts were refined following
Mondada’s (2018) conventions for embodied conduct (see Appendix).
The present study is based on a collection of 33 occurrences of self-increment-
ed RCs found in the data. We conducted sequential analysis of all occurrences so
as to identify their interactional workings. In what follows, we present the two
most consistent uses found in the data: Doing referential repair (7 occurrences)
(Section 4) and doing referential elaboration (13 occurrences) (Section 5). The
remaining occurrences comprise uses for fishing for recipient reaction (Stoenica
et al. to appear) and doing enumeration (Stoenica, 2016b, 2018).
4. RCs as self-increments – I: A resource for doing referential repair
In this section we discuss speakers’ use of self-incremented RCs for doing referen-
tial repair. In the data, these RCs occur in the following environment:
Speaker A produces a pragmatically and syntactically complete turn ending

on turn-final intonation, i.e., on a complex TRP.
Speaker B displays trouble with referent recognition,
Speaker A extends his or her turn via an RC.
A first illustration is provided in Excerpt (6), which reproduces – in more detail
and within a larger context – Excerpt (5) quoted above. The extract is taken from a
conversation between David and Gebbe (pictured on the left side of figure 1, l. 02):
(6) ‘Que Romain il nous a parlé’ [Corpus Pauscaf (Pause 18): 02m10-02m18]
01 DAV: j’ai regardé le film Safe.
‘I’ve watched the movie Safe’
02 ×*(0.6)*×
geb ×raises eyebrows and wrinkles brow×
*quick middle-distance look to left*
fig.1
03 DAV: que Romain il nous a parlé *une fois.=
PRO.REL-DO Romain he us AUX tell.PP DET time
‘that Romain once told us about’
geb *quick middle-
distance look
to left->l.04
04 GEB: =euh*avec [euh
‘with’
--->*looks at dav------------>>
05 DAV: [avec eu[:h

‘with’
06 GEB: [Statham?
07 DAV: Statham ouais=

‘Statham yeah’
08 GEB: ×=il est bien hein?×

‘it’s good huh’
×relaxes face, large smile×
fig.2
The RC (l.03) occurs as a post-gap increment (Schegloff, 2000, 2016). That is, it oc-
curs after a complex TRP that is followed by a gap. As mentioned above, an analysis
of the mere verbal conduct of the participants may suggest that the increment
here functions just like many other increments, providing a second opportunity
for recipient reaction: David, in line 01, offers an informing, which makes recipi-
ent response relevant (Thompson et al., 2015). However, such response appears
to be missing (l. 02).
Analysis of participants’ embodied conduct sheds a different light on the
excerpt. While the RC is produced after a noticeable absence of the recipient’s
verbal reaction, it follows the recipient’s non-verbal display of trouble (l. 02):
Gebbe’s raised eyebrows and “out of focus ‘middle-distance’ look” (Goodwin,
1987: p. 117) – extending over a 0.6 gap – display a thinking face (Goodwin and
Goodwin, 1986) that suggests cognitive search. Gebbe’s non-verbal conduct does
not specify the trouble source and can hence be seen as an open class repair initia-
tor (cf. Drew, 1997) that is delivered non-verbally: It is an embodied open class
repair initiator. David in turn treats Gebbe’s multimodal conduct as signaling a
potential problem with identifying the referent David had introduced by means of
the recognitional form (Sacks and Schegloff, 1979) le film Safe ‘the movie Safe’. By
adding the RC que Romain il nous a parlé une fois ‘that Romain once told us about’,
David provides further information potentially enhancing referent recognition, in
an attempt to solve the referential problem exhibited by Gebbe. This interpretation
is corroborated by the further course of the interaction. At line 04, Gebbe engages
in a word search activity regarding the name of the movie’s leading actor, and
thereby displays recognition of the movie in question. He then delivers a candidate
name (l. 06), which is confirmed by David (l. 07).
It is only at this moment of the interaction that the referential problem appears
to be solved and the repair sequence that started at line 02 is brought to a close.
This is evidenced by the fact that, in line 08, Gebbe responds to David’s initial in-
forming (l. 01), by positively assessing the movie in question and thereby claiming
access to the referent. This epistemic shift is further highlighted by Gebbe’s mul-
timodal conduct, his face relaxing and displaying a large smile, while delivering
the evaluative response. As noted by Thompson et al. (2015: p. 51), “[i]n general,
the recipient to an informing responds in such a way as to register an emerging

epistemic shift from K − to K +”. This type of response is exactly what was missing
in line 02, but is now delivered in line 08.
In light of the above, David’s self-incremented RC (l. 03) can be interpreted as
a way of doing referential repair while maximizing the progressivity of talk: It pro-
vides a referential specification that is syntactically continuous with the turn-so-
far, and hence minimizes disruptiveness, while repairing trouble with referent rec-
ognition that has been displayed through the co-participant’s embodied conduct.
Similar observations pertain to Excerpt (7), taken from the same interaction.
Gebbe and David continue talking about movies with the actor Jason Statham. At
line 07, in brief overlap with Gebbe, David starts to compare Statham with another
actor whose name he does not seem to recall. He explicitly calls for Gebbe’s help:
c’est le nouveau comment il s’appelle l’autre chauve? ‘it’s the new what’s the name of
the other bald (guy)?’.
(7) ’Qui joue dans the die hard’ [Corpus Pauscaf (Pause 18): 02m39-02m51]
01 GEB: [c’est une histoire géniale.
‘it’s a great story’
02 DAV: [°ok°
03 *∆(0.7)∆*
geb *gazes down smiling, plays with a coffee cream capsule-->l.05
dav ∆coffee stir stick in the mouth, looks at what geb is doing∆
04 GEB: Statham c’est,

‘Statham it’s’
05 ∆(1.5)*∆
geb ----->*
dav ∆takes coffee stir stick out of his mouth∆
06 GEB: [*c-*
*raises head and gazes at dav*
07 DAV: [c’est le nouveau comment il s’appelle l’autre

‘it’s the new what’s the name of the other
08 chauve?
bald (guy)’
09 *(1.1)*
geb *quick middle-distance look to his left, then gazes at dav*
10 DAV: ∆eu[:h∆
∆hits table with the coffee stir stick∆
11 GEB: [*°l’autre chau[ve°?

‘the other bald (guy)’
*questioning look at dav------>l.12
fig.1
12 DAV: [qui joue dans the die hard¿*

PRO.REL-SBJ plays in DET die hard
‘who plays in die hard’
--------------------------->*
13 *(0.7)*
geb *quick middle-distance look to his left*
fig.2
14 GEB: *ah Bruce Willis?=

*looks at dav, smiling, gently raising eyebrows and head--->>
fig.3
15 DAV: ∆=Bruce Willis.∆

∆nods-----------∆
In response to David’s question (l. 07–08), Gebbe remains silent, his embodied
conduct exhibiting doing thinking (l. 09): He stares into space, displaying a quick
middle-distance look to his left, and then again gazes at David as if inviting him
to pursue (on the role of gaze for turn-allocation and pursuing response, see
Sacks et al., 1974; Goodwin, 1979; Stivers and Rossano, 2010). He then explicitly
pinpoints the trouble source in David’s turn (l. 11) by repeating, with low volume
and slightly rising intonation, l’autre chauve? ‘the other bald guy?’, accompanied
by a questioning look at David. In an attempt to address the referential problem
conveyed by Gebbe’s both verbal and embodied conduct, David (l. 12) in turn
adds the RC qui joue dans the die hard¿ ‘who plays in die hard¿’. Given that it comes
in in overlap with Gebbe’s l’autre chauve? ‘the other bald guy’ (l. 11), David’s
RC appears to be produced as a self- rather than an other-increment. This self-
incremented RC is deployed by David as a resource for repairing the referential
problem. The slightly final rising intonation of the RC suggests that David calls
for a sign of reference recognition from Gebbe, which Gebbe does by means of
‘ah Bruce Willis’ (l. 14). Gebbe’s shift from displaying problems with the referent
and initiating referential repair to finally identifying the troublesome referent is
beautifully borne out not only by the use of the change-of-state token ah prefacing
the referent identification but also by his embodied conduct (l. 14): Gebbe looks
at David, smiling, gently raising his eyebrows and head. The sequence is closed by
David’s confirming repetition plus nodding (l. 15).
The same use of RC increments for doing referential repair can be observed
in the following excerpt, taken from a focus-group discussion that is organized by
Monica and that brings together four secondary school pupils. The excerpt begins
with Delphine explaining her drawing of a bilingual person’s brain.
(8) ‘Qu’il veut parler’ [Corpus CLA-FNRS(F): (12m55s-13m12s)]
01 DEL: il est un peu partagé en ↑deux le cer↑veau
‘it’s a little divided in two the brain’
02 et puis que:.hh qu’il y a un bout français
‘and then that that there is a French piece’
03 et puis un bout italien,
‘and then an Italian piece’
04 MON: ↑ouai::s
‘yeah’
05 DEL: puis quand il parle il arrive à retrouver
‘then when he speaks he manages to find’
06 la langue.
‘the language’
07 (1.0)
08 DEL: °qu’ il veut parler quoi°.=
pro.rel-do he wants speak-inf prt
‘that he wants to speak prt.’
09 MON: =alors attends quand il parle donc la personne
‘so wait when he speaks so the person’
10 quand elle parle,
‘when she speaks’
11 (06) s- le cerveau arrive à retrouver la langue.
‘the brain manages to find the language’
12 DEL: ouais
‘yeah’
13 MON: c’est ça.
‘that’s it’
14 (1.4)
15 MON: d’accord ok.

‘all right ok’
Delphine’s explanation of the drawing, in lines 01, 02, 03, 05 and 06, is produced
after the turns of two of her fellow pupils, who have each described their own
drawing. Each of these two previous turns was met with an agreement token
and a positive assessment from Monica. Yet, here there is no such reaction from
Monica, although Delphine’s turn, at the end of line 06, ends on a complex TRP.
Following Monica’s lack of uptake (see the pause at l. 07), Delphine increments
her turn with the RC qu’il veut parler quoi ‘that he wants to speak’ (l. 08). This
suggests that Delphine is treating Monica’s lack of uptake as signaling her failure
to properly identify la langue ‘the language’ that Delphine has referred to in line
06. In an attempt to repair this problem of referential identification, Delphine of-
fers a specification of la langue ‘the language’ by means of the incremented RC.
This seems to successfully repair Monica’s problem, as the latter immediately seeks
confirmation (l. 09–11) of her own understanding of Delphine’s turn and, after
Delphine’s confirmation (l. 12), produces multiple signs of agreement (l. 13 and
15). These constitute the relevant next actions after Delphine’s description ending
at line 06, according to the routinized sequencing of the previous conversational
actions.
In this section, we analyzed the use of self-incremented RCs as a resource for
accomplishing referential repair. The quoted cases illustrate a consistent pattern
found with the 7 instances of self-incremented RCs doing referential repair in
the data: The incrementally added-on RC occurs as a response to a non-verbal
display of trouble (which may be accompanied by verbal display, see Excerpt 7)
on the part of the recipient upon a mention of some referent. The occurrence of
self-incremented RCs for accomplishing other-initiated referential repair does not
seem to be bound to a precise action environment: These RCs are found in the
data ensuing from informings (Excerpt 6), requests for information (Excerpt 7),
assessments, and assertions.
Clearly, the linguistic material that these increments are made of matters for
how they function interactionally: RCs, because they relate to a referent denoted
by some NP in the preceding clause, lend themselves perfectly well to doing ref-
erential repair by adding information on that previously mentioned NP. But what
do RC increments allow speakers to do that other ways of self-repairing reference
in turn-extensions (e.g., by means of a syntactically independent clause) would
not allow them to do? The analysis offered in this section points to the fact that
the unique contribution of RC increments is that they allow speakers to balance
the competing principles of intersubjectivity (maintaining mutual understanding)
and of progressivity (fostering the moving forward of the interaction). Heritage
(2007: p. 260) argues that the participants’ dealing with referential issues basically
creates a conflict between these two principles: Referential repair involves the
principle of intersubjectivity (establishing referential common ground) interfering
with the principle of progressivity, as it momentarily suspends the moving forward
of talk. Crucially, the incremental adding of an RC to a turn-so-far provides one
practical solution for favoring referent recognition while minimizing the disruption
of repair by presenting the repair segment as syntactically continuous with the stretch
of talk containing the repairable. RC self-incrementation therefore represents a
grammatical practice that allows speakers to maximize the compatibility between
the principles of intersubjectivity and of progressivity (see Pekarek Doehler, 2011
and Horlacher, 2015 for how right-dislocations can be used in related ways).
5. RCs as self-increments – II: A resource for doing referential elaboration
In this section we analyze a second interactional use of self-incremented RCs,

which differs from the use documented above in terms of the contingencies it
responds to and the interactional consequentialities it entails. In the data, these
RCs occur in the following environment:
Speaker A produces a pragmatically and syntactically complete turn ending
on a try-marked recognitional referent (Sacks and Schegloff,
1979).
Speaker B produces an “assertion of recognition” (Sacks and Schegloff,
1979: p. 19) of the referent (‘mhm’, ‘yeah’, etc.),
Speaker A extends his or her turn via a RC.
This type of RC hence occurs after a turn that has reached a TRP marked by
syntactic and local pragmatic completion (Ford and Thompson, 1996), but that
ends on a try-marked intonation (Sacks and Schegloff, 1979). In the literature,
increments have been defined as following turns ending on final falling or rising
intonation (see Section 4); syntactic extensions of turns ending on try-marking
have not yet been discussed. Such turns ending on try-marking intonation are
treated by participants as accomplishing a distinct action, namely eliciting refer-
ent recognition by the recipient (see Sacks and Schegloff, 1979). It is based on
the turn’s being treated as complete with regard to this distinct practical purpose
that we interpret the added-on RC as an increment. As shown in the schematic
representation above, the RCs under scrutiny systematically occur as post-other-
talk increments (Schegloff, 2000, 2016), the talk in this case consisting merely
of a “small, non-floor-taking turn” (Ford and Thompson, 1996: p. 150). Rather
than accomplishing referential repair, the self-incremented RC in this case of-
fers referential elaboration: The speaker provides further information on behalf
of a referent after having received confirmation of referent recognition on the

part of the recipient.
Excerpt (9) is taken from an audio-recorded focus-group discussion with
teacher trainers. Barbara reports on one of her colleagues, a teacher of Spanish,
who has only one intern to supervise, and consequently has enough time for
teaching (rather than training interns).
(9) ‘La maîtresse de méthodologie qui’ [Corpus FNRS (A): 01m24-01m34]
01 BAR: nous avons la maîtresse d’espa- la m-
‘we have the teacher of Spa- the t-‘
02 la maî- la maîtresse de méthodologie d’espagnol¿
‘the tea- the teacher of Spanish methodology’
03 MON: ouais
‘yeah’
04 BAR: .h qui a UN stagiaire,
pro.rel-sbj has one intern
‘who has one intern’
05 [.h ] c’est clair qu’elle enseigne ↑plus,
‘it’s obvious that she teaches more’
06 MON: [mhm]
07 BAR: (0.8) que:[: qu’elle ne FORME dans ce cas-là.]
‘than she provides training in that case’
08 MON: [bien sûr °ouais ouais ouais (0.2) ]&
‘of course yeah yeah yeah’
09 &ouais°.
‘yeah’
Barbara’s turn at lines 01 and 02, which will become extended by means of an
RC increment (l. 04), presents a potentially syntactically complete unit, yet does
not end on a complex TRP. Rather, it ends on a slightly rising intonation that can
be heard as marking the newly introduced referent la maîtresse de méthodologie
d’espagnol, ‘the teacher of Spanish methodology’ as a try (Sacks and Schegloff,
1979), thereby soliciting recipient’s display of referent recognition. Sacks and
Schegloff (1979) have identified try-marking as a routine procedure for securing
referential common ground: A participant produces a recognitional form (e.g., a
proper or a common noun) with a rising intonation, and then pauses momentarily,
thereby making co-participant’s confirmation of the referent relevant as a next.
This is exactly what happens in Excerpt (9), with the particular characteristic that
the try-marking occurs at the end of a syntactically and pragmatically complete
unit.3 The try-marking is responded to by Monica by a ouais ‘yeah’ (l. 03), which,
suggests that Monica does not encounter any trouble with referent recognition
3. The turn-so-far is pragmatically complete in as far as it can be heard as an informing. It

reaches a point of “local pragmatic completion” as defined by Ford and Thompson (1996: p. 150).
and thus it functions as a go-ahead signal. It is only after having secured mutual
common ground in this way that Barbara adds the incremental RC qui a UN stagi-
aire ‘who has one intern’ (l. 04) and further continues her turn.
As opposed to what we have observed in Section 4 above, the self-incremented
RC does not respond to a recipient’s display of trouble but occurs after the re-
cipient’s display of referent recognition. Rather than providing further elements in
view of referent identification, it adds information on behalf of an already ratified
referent. Also, the production of the RC entails other interactional consequentiali-
ties. While the ‘repair’ type of RC increment ultimately leads up to a recipient’s
display of shift from K- to K+ status, no such epistemic claims are observed fol-
lowing the type of RC increment illustrated in Excerpt (9).
Yet, again, the bi-clausal pattern is configured on-line for the purpose of dealing
with local interactional needs and affordances, namely securing common ground,
before further elaborating on a given referent: What is first produced as a simple
SVO pattern, ‘we have the teacher of Spanish methodology’ (l. 01–02), emerges in
the end as a presentational construction of the ‘we have X that/who’-type.
The following excerpt shows a similar case in point. The excerpt stems from
a focus-group discussion, organized by Céline, that brings together four teachers.
The participants have been talking about how one may become bilingual, and
Céline gives the example of the Bulgarian-born essayist Elias Canetti, from whose
writings she chose an excerpt and who became bilingual out of love for his mother.
(10) ‘Qui en fait faisait l’effort’ [Corpus FNRS (I): 47m30-47m51]
01 CEL: eu::h justement il y avait beaucoup de documents,
‘u::m precisely there were many documents’
02 .h mais en fait on avait sélectionné un ex- un
‘but in fact we had selected an ex- an’
03 extrait de Cane- Canetti¿
‘excerpt from Cane- Canetti’
04 (0.3)
05 LOU: mhm
06 (0.2)
07 CEL: qui en fait faisait l’effort d’apprendre
pro.rel-sbj in fact made det effort prep learn-inf
‘who in fact made the effort to learn’
08 une langue finalement par amour de sa mère.
det language finally prep love of his mother
‘a language after all out of love for his mother’
09 ((turn continues))
Just like in Excerpt (9), the speaker does not continue talking about the target
referent (here: Canetti) until she receives an acknowledgement token (here: mhm,
l. 05) from her co-participant – a reaction that she calls for both by the try-marked
intonation with which she produces the name ‘Canetti’ (l. 03), and by her refraining
from talking after having produced that name, as indicated by the slight pause at
line 04. By producing the acknowledgement token mhm (l. 05), Louis confirms that
he treats Céline’s pausing at line 04 and her prosodic delivery of Canetti as eliciting
a reaction on his part. It is only after the production of this acknowledgement that
Céline uses an RC (l. 07–08) in order to elaborate on the newly introduced referen-
tial expression. The production of the RC is thus contingent on the co-participant’s
reaction and it emerges as a result of an interactionally established reference.
Excerpt (11) provides a further illustration. Here, the speaker’s embodied
conduct further corroborates the interpretation offered so far. Penny (on the left
side of the picture) and Danielle are two students sharing a coffee break. Penny
reports on an assignment given by a teacher (‘he’, l. 01) that she found unclear.
(11) ‘Des données’ [Corpus Pauscaf (Pause 9): 22m12-22m19]
01 PEN: *∆donc il a dit qu’il nous fournissait des données¿
‘so he said that he would provide us with data’
*gazes at dan----------------------------------------------->l.02
dan ∆gazes at pen----------------------------------------------->>
02 (0.5)*
pen ---->*
03 DAN: *ouais
‘yeah’
pen *looks down--------------------------------------------->l.04
04 PEN: ((lip smack)) alors qu’ on allait-*

PRT PRO.REL-DO PRO.INDEF.3SG was going
‘PRT which we would’
pen ----------------------------------------------->*
05 *qu’ on analyserait à deux

PRO.REL-DO PRO.INDEF.3SG analyze.PCON PREP two
‘which we would analyze (in pairs of) two’
pen *gazes at dan------------------------------------------------>>
06 pour le mois de no↑vembre,

PREP DET month of November
‘for the month of November’
07 et pis après tous seuls pour le mois de décembre.

and then after all alone PREP DET month of December
‘and later all alone for the month of December’
08 mais du coup je suis allée voir

‘but so I went to see’((the turn continues))
Just like in the preceding excerpts, the incrementally added RC occurs after co-
participant’s display of referent identification (l. 03) ensuing from the try-marked
referent (l. 01). The subsequent pause (l. 02) following the try-marked referent
may suggest that a recipient reaction is due; this is further corroborated by the
speaker’s maintenance of gaze on the recipient (cf. Stivers and Rossano, 2010). The
recipient’s (Danielle’s) response in line 03 shows that she treats the try-marking as
a try-marking: She provides a go-ahead (which does not further claim the floor),
and thereby indicates that there is no trouble with referent recognition, and also,
that she herself is not claiming the floor. This is again corroborated by Danielle’s
maintenance of gaze on Penny from line 2 on, which embodies her expectancy
for Penny to continue.
Of central importance for understanding how the two clauses in the [main
clause + RC] pattern are combined and become interactionally functional is the
prosodic delivery of the first clause, conjointly with co-occurring gaze conduct.
Contrary to the ‘repair’ type of RC increment, where the first clause ends on a
complex TRP and the turn extension is occasioned by recipient’s display of trouble,
with the referential elaboration type of RC increment, the first clause ends on a
mid-rise of pitch, typical for try-marking, which can be heard as calling for a sign
of recipiency (e.g., by means of a response token) while still projecting more to
come on the part of the speaker, that is, without yielding the floor. The try-marking,
rather than signaling immediate continuation, signals a potential continuation of
the first speaker’s turn after the recipient’s response token, and this is convergent
with the recipient’s maintaining gaze on the speaker, as illustrated in Excerpt (11).
Thus, a possible turn extension is projected past the recipient’s response, yet the
grammatical trajectory as well as the action that is to be accomplished by that
extension remain unspecified.
In light of the features illustrated in Excerpts (9) to (11), we might again ask
why speakers resort to complex syntax, that is why they choose to offer further
elaboration on a referent by means of a post-other talk incremented RC, rather
than, for example, by starting a new turn with a simple clause that is not marked as
syntactically related to preceding talk. Issues of progressivity are again at play here.
By means of the self-incremented RC, the speaker embeds the checking of referent
recognition within one single syntactic (and turn-) trajectory. The RC hence serves
as a resource for tying back to the respective referent in the speaker’s preceding
turn, skipping over the interactional sequence comprising the speaker’s call for
referent recognition and the recipient’s confirmation, and thereby maximizing the
progressivity of talk. The checking of referent recognition, its confirmation, and
further predication on that referent are thereby smoothly integrated into a larger
trajectory of a single turn and its extension (for related uses of on-line emergent
left-dislocations see Pekarek Doehler, 2011 and Pekarek Doehler et al., 2015).
6. Discussion
In this paper we set out to analyze how the clause-combining pattern of the type
[main clause + relative clause] – a key exemplar of complex syntax – emerges in
use ‘on the fly’, as part of speakers’ dealing with local interactional contingencies.
While we lack evidence enabling us to discuss whether interactional organization
may motivate the very existence of clause-combining patterns in the language, we
do have evidence for the fact that during the real-time production of turns and
actions, speakers patch together these patterns ad hoc to get precise interactional
jobs accomplished, such as establishing referential common ground and ulti-
mately maintaining intersubjectivity. These findings extend our understanding of
on-line emergence (Hopper, 1987, 2011; Auer, 2009; Deppermann and Günthner,
2015; Pekarek Doehler et al., 2015) beyond phrasal or clausal constructions to the
concatenation of syntactically related clauses, and hence to complex syntax.
We have identified two recurrent uses of self-incremented RCs in response
to two distinct interactional contingencies. In response to a recipient’s verbal or
nonverbal display of trouble, speakers produce incremented RCs to accomplish
referential repair; these RCs are not projected (they follow a complex TRP) but
are triggered by recipient verbal or non-verbal conduct. By contrast, in response
to recipients’ display of referent recognition following a try-marking of that
referent, speakers use RCs to offer further information about that referent, that
is, they elaborate on a referent once its recognizability has been interactionally
established. This second use of RC increments is therefore not doing repair, yet
the try-marking that precedes it can be heard as designed to pre-empt potential
trouble with referent recognition. In both cases, the clause-combining pattern of
the type [main clause + relative clause] emerges in real time for dealing with local
interactional needs related to the securing of referent recognition while at the same
time maximizing the progressivity of talk: It allows speakers to strike a balance
between the competing principles of intersubjectivity (mutual understanding) and
progressivity (cf. Heritage, 2007).
The way the [main clause + relative clause] pattern is patched together on the
fly has consequences for our understanding of the very nature of the RC. Within
the real-time movement from turn to turn, the nature of the RC (traditionally
labeled as restrictive vs. non-restrictive) is interactionally determined, rather than
based on semantic-pragmatic properties inherent in the lexico-syntactic make-up
of the clause-combining pattern. We have provided two types of evidence in this
regard. On the one hand, RCs that are produced after referent recognition has been
interactionally secured (Section 5) cannot be, from the participants’ perspective,
anything but non-restrictive in nature, since they do not contribute to the referen-
tial identification of their head NPs. On the other hand, the incremented RCs that
are used for doing referential repair (Section 4) seem restrictive in nature, exactly
because they are deployed for the purpose of referent identification by recipients
after recipients’ display of initial trouble with such identification. Importantly,
these repairing RC increments are in the data sometimes added after the speaker’s
mention of proper nouns (see Excerpt 6). Yet, according to the traditional litera-
ture on RCs, proper names are considered to be referentially self-sufficient, and
RCs relating to proper names as their head NPs are categorized as non-restrictive
because they are understood as not contributing to referential identification
(Givón, 1990: p. 650; Le Goffic, 1994). Our analysis, however, has shown quite
the contrary: Speakers do mobilize RCs after proper names precisely for securing
referential identification – a function classically attributed to restrictive RCs –,
and this is done in order to address recipients’ trouble with the recognition of
such proper nouns. These findings call into question classical perspectives on the
functioning of RCs as well as abstract, decontextualized views of the semantic-
pragmatic nature of these grammatical constructions.
Furthermore, because they can be added after a recognizable clause-ending,
RCs provide ideal material for incrementation. The findings reported in this
paper add to a growing number of empirical studies evidencing that the concrete
linguistic material that increments are composed of matters for the interactional
purposes they fulfill (Clift, 2007; Seppänen and Laury, 2007; Horlacher, 2015).
Technically speaking, any increment creates a second opportunity for recipient
reaction, as it leads up to a second TRP. Yet, the precise interactional purpose it
fulfills beyond this generic feature, as well as the kind of response it makes relevant,
hinges not only on the sequential location of the increment, but also on what lin-
guistic material it is made of. This is a point that deserves to be more systematically
investigated in future research in view of a more differentiated understanding of
the interactional workings of increments.
7. Conclusion
Through the present study, we hope to have contributed to an understanding of

the use of RCs from a grammar-in-interaction perspective. The findings can be
brought to bear on three much-debated issues.
First, the excerpts quoted in this paper illustrate how the use of RCs may
emerge from interactional contingencies rather than being motivated by semantic
referential constraints inherent in the lexico-syntactic make-up of the construc-

tion. In particular, in light of the findings, the restrictive vs. non-restrictive nature
of RCs emerges as a local interactional accomplishment, contingently negotiated
among participants. This latter point radically inverses the traditional understand-
ing of the semantic-pragmatic nature of the RCs, showing how that very nature
arises through real-time production in situated language use (cf. Stoenica 2018 for
an in-depth discussion of this issue).
Second, the inherently subordinate status of the RCs appears to be seriously
challenged by the use of RCs as increments. In particular, the analysis of RC in-
crements doing referential repair (Section 4) has shown that incremented RCs
are treated both by speakers and by recipients as interactionally relevant units of
talk that accomplish a precise action in its own right and with its own sequential
implications. We have provided further evidence in support of the praxeological
autonomy of self- as well as other-incremented RCs in Stoenica (2016a, 2018) and
Stoenica et al. (to appear), evidencing that these can accomplish a range of base ac-
tions, such as doing disagreement or initiating repair. The fact that incrementally
added RCs often stand as independent prosodic units that are not prosodically
projected by preceding talk further supports this argument (see Stoenica et al.
to appear for detailed prosodic analyses). Taken together, these observations
highlight the fact that what formally appear as RCs may in fact be used as au-
tonomous units of talk and action, triggered by local interactional contingencies.
Accordingly, the interactional use of RCs may not be restricted to a single syntactic
relation, that of subordination, further suggesting that syntactic independence vs.
subordination may range on a continuum (Foley and Van Valin, 1984; Matthiesen
and Thompson, 1988; Günthner, 1996; Auer, 1998; Evans, 2007) and, ultimately,
that some occurrences of RCs are not part of clause-combining patterns at all (cf.
Béguelin et al., 2000), and hence may be cases of insubordination (Evans, 2007, see
also Maschler, this volume).
Third, the use of RC increments as praxeologically autonomous units of talk
goes against the classical understanding of increments as doing continuation of the
actions accomplished in the host-turns (Schegloff, 1996; see Section 2 above). The
data discussed in this paper show that RC increments can accomplish an action
in its own right, that is, an action that is distinct from the action accomplished by
the host-turn. Such increments may hence best be understood as representing new
TCUs rather than TCU continuations, despite the fact that their surface marking
(the presence of a relative pronoun) displays them as syntactically continuous with
the preceding turn. This latter point emphasizes, once again, the importance of the
linguistic material increments are made of for their precise interactional workings
and foregrounds the need for a further specification of the notion of increment in
the light of both its lexico-syntactic composition and its praxeological potential.
Finally, we hope to have provided further empirical evidence testifying to the

thoroughly temporal and emergent nature of grammar in use, by extending the
on-line perspective to issues of complex syntax. The findings suggest that clause-
combining in talk-in-interaction is a contingent product of how people deal with
practical problems related to coordinating social interaction, warranting its pro-
gressivity and maintaining intersubjectivity. The findings also show that a fuller
understanding of the working of clause-combining in interaction – and more
generally of grammar-in-interaction – calls for close analyses of how participants’
verbal conduct interfaces with embodied conduct within the complex local ecol-
ogy of social interaction (Goodwin, 1979; Keevallik, 2018). The grammar-body
interface is another path that future research on grammar-in-interaction, action
formation, and action ascription will need to venture on.
Glosses
aux auxiliary pro.indef indefinite pronoun

det determiner pro.rel relative pronoun
do direct object prt particle
inf infinitive sbj subject
pcon present conditional 3sg third person singular
pp past participle
References
Auer, P. (1998). Zwischen Parataxe und Hypotaxe: ‘abhängige Hauptsätze’ im Gesprochenen

und Geschriebenen Deutsch. Zeitschrift für Germanistische Linguistik 26, 284–307.
https://doi.org/10.1515/zfgl.1998.26.3.284
Auer, P. (2007). Why are increments such elusive objects? An afterthought. Pragmatics 17,
647–658. https://doi.org/10.1075/prag.17.4.03aue
Auer, P. (2009). On-line syntax: Thoughts on the temporality of spoken language. Language
Béguelin, M.-J., Matthey, M., Bronckart, J.-P. & Canelas-Trevisi, S. (2000). De la phrase aux
énoncés : grammaire scolaire et descriptions linguistiques. Bruxelles: De Boeck Duculot.
Berrendonner, A. (2008). L’alternance que / #. Subordination sans marqueur ou structure
périodique? In D. Van Raemdonck (Ed.), Modèles syntaxiques (pp. 279–298). Berne: Lang.
Berrendonner, A. & [Reichler-]Béguelin, M.-J. (1989). Décalages : les niveaux de l’analyse
linguistique. Langue Française 81, 99–125. https://doi.org/10.3406/lfr.1989.4770
Brunner, J.-J. (1981). Ces relatives qui n’en sont pas. L’information grammaticale 8, 12–16.
https://doi.org/10.3406/igram.1981.2445
Clift, R. (2007). Grammar in time: the non-restrictive ‘which’-clause as an interactional resource.

Essex Research Reports in Linguistics 55, 51–82.
Couper-Kuhlen, E. (1996). Intonation and clause-combining in discourse: the case of because.
Pragmatics 6, 389–426. https://doi.org/10.1075/prag.6.3.04cou
Couper-Kuhlen, E. (2012). Turn continuation and clause combinations. Discourse processes 49,
273–299. https://doi.org/10.1080/0163853X.2012.664111
Couper-Kuhlen, E. & Ono, T. (Eds.) (2007). Turn continuation in cross-linguistic perspective.
Pragmatics 17.
Deppermann, A. & Günthner, S. (Eds.) (2015). Temporality in Interaction. Amsterdam/Philadel-
phia: John Benjamins. https://doi.org/10.1075/slsi.27
Deulofeu, J. (1999). Questions de méthode dans la description morphosyntaxique de l’élément
QUE en français contemporain. Recherches sur le français parlé 15, 163–198.
Drew, P. (1997). ‘Open’ class repair initiators in response to sequential sources of troubles in
conversation. Journal of Pragmatics 28, 69–101.
https://doi.org/10.1016/S0378-2166(97)89759-7
Evans, N. (2007). Insubordination and its uses. In I. Nikolaeva (Ed.), Finiteness: Theoretical and
Empirical Foundations (pp. 366–431). Oxford: Oxford University Press.
Foley, W. A. & R. D. Van Valin Jr. (1984). Functional Syntax and Universal Grammar. Cambridge:
Cambridge University Press.
Ford, C. E. & Thompson, S. A. (1996). Interactional units in conversation: syntactic, intona-
tional, and pragmatic resources for the management of turns. In E. Ochs, E. A. Schegloff &
S. A. Thompson (Eds.), Interaction and Grammar (pp. 134–185). Cambridge: Cambridge
Ford, C. E., Fox, B. A. & Thompson, S. A. (2002). Constituency and the grammar of turn in-
crements. In C. E. Ford, B. A. Fox & S. A. Thompson (Eds.), The Language of Turn and
Sequence (pp. 14–38). Oxford: Oxford University Press.
Fox, B. A. & Thompson, S. A. (2007). Relative clauses in English conversation. Relativizers,
frequency, and the notion of construction. Studies in Language 31, 293–326.
https://doi.org/10.1075/sl.31.2.03fox
Gapany, J. (2004). Formes et fonctions des relatives en français. Étude syntaxique et sémantique.
Berne: Peter Lang.
Givón, T. (1990). Syntax. A Functional-typological Introduction. Vol. II, Amsterdam: John Ben-
jamins.
Givon, T. (1995). Functionalism and Grammar. Amsterdam/Philadelphia: John Benjamins.
https://doi.org/10.1075/z.74
Goodwin, Ch. (1979). The interactive construction of a sentence in natural conversation. In G.
Psathas (Ed.), Everyday Language: Studies in Ethnomethodology (pp. 97–121). New York:
Irvington.
Goodwin, Ch. (1987). Forgetfulness as an Interactive Resource. Social Psychology Quarterly 50,
115–131. (Special Issue on Language and Society, edited by Douglas Maynard).
Goodwin, M. H. & Goodwin, Ch. (1986). Gesture and coparticipation in the activity of search-
ing for a word. Semiotica 62, 51–75.
Groupe de Fribourg (2012). Grammaire de la période. Berne: Peter Lang.
Günthner, S. (1996). From Subordination to Coordination? Verb-Second Position in German
Causal and Concessive Constructions. Pragmatics 6, 323–371.
https://doi.org/10.1075/prag.6.3.05gun
Günthner, S. (2008). Die Sache ist …’: eine Projektorkonstruktion im gesprochenen Deutsch.
Zeitschrift für Sprachwissenschaft 27, 39–72. https://doi.org/10.1515/ZFSW.2008.003
Heritage, J. (2007). Intersubjectivity and progressivity in person (and place) reference. In N. J.
Enfield & T. Stivers (Eds.), Person Reference in Interaction (pp. 255–280). Cambridge:
Cambridge University Press.
Hopper, P. (1987). Emergent grammar. Proceedings of the Thirteenth Annual Meeting of the
Berkeley Linguistics Society, 139–157.
Hopper, P. (2011). Emergent grammar and temporality in interactional linguistics. In P. Auer
& S. Pfänder (Eds.), Constructions: Emerging and Emergent (pp. 22–45). Berlin/Boston:
Walter de Gruyter. https://doi.org/10.1515/9783110229080.22
Hopper, P. & Thompson, S. A. (2008). Projectability and clause combining in interaction. In R.
Laury (Ed.), Crosslinguistic Studies of Clause Combining. The Multifunctionality of Conjunc-
tions (pp. 99–124) Amsterdam: John Benjamins. https://doi.org/10.1075/tsl.80.06hop
Horlacher, A.-S. (2015). La dislocation à droite revisitée. Une approche interactionniste. Louvain-
la-Neuve : De Boeck. https://doi.org/10.3917/dbu.hcher.2015.01
Horlacher, A.-S. & Pekarek Doehler, S. (2014). ‘Pivotage’ in French talk-in-interaction: on the
emergent nature of [clause-NP-clause] pivots. Pragmatics 24, 593–622.
https://doi.org/10.1075/prag.24.3.07hor
Jespersen, O. (2006 (1993)). Essentials of English grammar. London: G. Allen and Unwin. (con-
sulted in the version Routledge/Taylor & Francis e-Library, 2006).
Kärkkäinen, E. (2003). Epistemic Stance in English conversation: A Description of its Interactional
Functions, with a Focus on I think. Amsterdam: John Benjamins.
on Language and Social Interaction 46, 1–21. https://doi.org/10.1080/08351813.2013.753710
Keevallik, L. (2018). What does embodied interaction tell us about grammar? Research on
Laury, R. (Ed.) (2008). Crosslinguistic Studies of Clause Combining. The Multifunctionality of
Conjunctions. Amsterdam/Philadelphia: John Benjamins. https://doi.org/10.1075/tsl.80
Laury, R. & Suzuki, R. (Eds.) (2011). Subordination in Conversation. A cross-linguistic perspec-
tive. Amsterdam/Philadelphia: John Benjamins. https://doi.org/10.1075/slsi.24
Laury, R. & Helasvuo, M.-L. (2015). Detached NPs with relative clauses in Finnish conversa-
tions. In M. M. J. Fernandez-Vest & R. D. Van Valin Jr. (Eds.), Information Structuring
of Spoken Language from a Crosslinguistic Perspective (pp. 149–166). Berlin/Boston: De
Gruyter Mouton. https://doi.org/10.1515/9783110368758-008
Le Goffic, P. (1994). Indéfinis, interrogatifs, relatifs (termes en QU-) : parcours avec ou sans
issue. Faits de langues 2, 31–40. https://doi.org/10.3406/flang.1994.935
Lindström, J. (2006). Grammar in the service of interaction: Exploring turn organization in
Swedish. Research on Language and Social Interaction 39, 81–117.
Lindström, J., Maschler, Y. & Pekarek Doehler, S. (Eds.) (2016). Grammar and negative epis-
temics in talk-in-interaction: Cross-linguistic studies. Journal of Pragmatics, 106.
Matthiesen, C. & Thompson, S. A. (1988). The structure of discourse and subordination. In
Haiman, J. & Thompson, S. A. (Eds.), Clause Combining in Grammar and Discourse. (pp.
275–329). Amsterdam: John Benjamins.
Maschler, Y. (2011). On the emergence of adverbial connectives from Hebrew relative clause
constructions. In P. Auer & S. Pfänder (Eds.), Constructions: Emerging and Emergent
(pp. 293–331). Berlin/Boston: De Gruyter. https://doi.org/10.1515/9783110229080.293
Mondada, L. (2018). Conventions for multimodal transcription. Retrieved from: https://franzo-
esistik.philhist.unibas.ch/fileadmin/user_upload/franzoesistik/home/Personen/Mondada/
Unterordner/Mondada_conv_multimodality.pdf
Norén, N. (2013). Pivots constructions as methods for perspective shift during turns at talk.
Pekarek Doehler, S. (2011). Emergent grammar for all practical purposes: The on-line format-
ting of dislocated constructions in French conversation. In P. Auer & S. Pfänder (Eds.),
Constructions: Emerging and Emergent (pp. 45–88). Berlin: Mouton de Gruyter.
https://doi.org/10.1515/9783110229080.45
Pekarek Doehler, S. & Horlacher, A.-S. (2013). The patching together of pivot-patterns in talk-
in-interaction: On ‘double dislocations’ in French. Journal of Pragmatics 54, 92–108.
Pekarek Doehler, S., De Stefani, E. & Horlacher, A.-S. (2015). Time and Emergence in Grammar.
Left-dislocation, Right-dislocation, Topicalization, and Hanging Topic in French Talk-in-
interaction. Amsterdam: Benjamins. https://doi.org/10.1075/slsi.28
Riegel, M., Pellat, J.-C. & Rioul, R. (2016). Grammaire méthodique du français. 6e éd., Paris: PUF.
Sacks, H., Schegloff, E. A. & Jefferson, G. (1974). A simplest systematics for the organization of
turn-taking for conversation. Language 50, 696–735. https://doi.org/10.1353/lan.1974.0010
Sacks, H. & Schegloff, E. A. (1979). Two preferences in the organization of reference to persons
and their interaction. In G. Psathas (Ed.), Everyday Language: Studies in Ethnomethodology
(pp. 15–21), New York: Irvington Publishers.
Schegloff, E. A. (1996). Turn organization: one intersection of grammar and interaction. In E.
Ochs, E. A. Schegloff & S. A. Thompson (Eds.), Interaction and grammar (52–134). Cam-
Schegloff, E. A. (2000). On turns’ possible completion, more or less: increments and trail-offs.
Paper delivered at the 1st Euroconference on Interactional Linguistics, Spa, Belgium.
Schegloff, E. A. (2016), Increments. In J. D. Robinson (Ed.), Accountability in Social Interaction,
Oxford: Oxford University Press. 239–263.
https://doi.org/10.1093/acprof:oso/9780190210557.003.0008
Seppänen, E.-L. & Laury, R. (2007). Complement clauses as turn continuations: The Finnish
et(tä)-clause. Pragmatics (Special Issue: Turn continuation in cross-linguistic perspective) 17,
553–572.
Stivers, T. & Rossano, F. (2010). Mobilizing response. Research on Language and Social Interac-
tion 43, 3–31. https://doi.org/10.1080/08351810903471258
Stoenica, I.-M. (2014). Répétition et différenciation dans les reprises structurelles intégrant des
relatives. In S. Rezzonico (Ed.), La parole reprise: formes, processus et fonctions. Actes du 12e
colloque de logopédie – 16–17 novembre 2012, TRANEL 60, 209–220.
Stoenica, I.-M. (2016a). Grammaire-en-interaction: le potentiel praxéologique des relatives
dans les conversations en français. Bulletin suisse de linguistique appliquée 104, 87–103.
Stoenica, I.-M. (2016b). Les relatives en contexte énumératif: une étude séquentielle. In O.
Galatanu, A.-M. Cozma & A. Bellachhab (Eds.), Représentations du sens linguistique. Les
interfaces de la complexité (pp. 257–270). Bruxelles : Peter Lang.
Stoenica, I.-M. (2018). Le rôle des relatives dans l’organisation séquentielle de l’interaction:
une approche temporelle, interactionnelle et praxéologique. [The role of relative clauses in
the sequential organisation of the interaction: A temporal, interactional and praxeological
approach]. Doctoral dissertation, University of Neuchâtel, Switzerland.
Stoenica, I.-M., Pekarek Doehler, S. & Horlacher, A.-S. (to appear). Emergent complex noun
phrases: On-line trajectories of ‘relativized’ NPs in French talk-in-interaction. In S. A.
Thompson & Y. Ono (Eds.), The pragmatics of the ‘noun phrase’ across languages: An emer-
gent unit in interaction, Amsterdam: John Benjamins, Typological Studies in Language series.
Tanaka, H. (1999). Turn-Taking in Japanese conversation: A study in grammar and interaction.
Tao, H. & McCarthy, M. J. (2001). Understanding non-restrictive which-clauses in spoken
English, which is not an easy thing. Language Sciences 23, 651–677.
https://doi.org/10.1016/S0388-0001(00)00026-7
Thompson, S. A. (2002). ‘Object complements’ and conversation. Towards a realistic account.
Thompson, S. A., Fox, B. A. & Couper-Kuhlen, E. (2015). Grammar in Everyday Talk: Building
Responsive Actions. Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9781139381154
Walker, G. (2001). A phonetic approach to talk-in-interaction. Increments in conversation. M.A.
thesis, University of York, UK.
Appendix
Transcription conventions for verbal conduct
[ start of overlap
] end of overlap
= latching (no pause, no overlap)
(0.7) measured pause in seconds and tenths of seconds
wo- truncated word
wo:rd syllable lengthening
? rising final intonation
¿ mid-rise intonation
. falling final intonation
, continuing intonation
word accentuation
°word° softer than surrounding speech
WORD louder than surrounding speech
↑word marked high rise in pitch (refers to the next syllable)
.h in-breath
((lip smack)) transcriber’s comment
Transcription conventions for embodied conduct

For more details see Mondada 2018
* * Indicates start and end of gaze or of gaze and other embodied

conduct (e.g., nodding) of speaker A.
× × Indicates start and end of an additional embodied conduct (e.g.,
smile) of speaker A.
∆ ∆ Indicates start and end of an embodied conduct of speaker B.
*---->l.12 Continuation of the described embodied conduct until line 12 of
transcript.
------->* End of the described embodied conduct
*----->> Continuation of the described embodied conduct until end of
excerpt.
Chapter 12
Afterword
Paul J. Hopper
Clauses and the emergent syntax of clause combining lie at the heart of the interac-
tional linguistics enterprise. They form the crucial link between the interpersonal
sociology of conversation and the standard concerns of the grammarian. The ar-
ticles in this book have illustrated the ways that, in various languages, speakers
deploy clauses and clausal fragments as they manage relationships with other
speakers and with the material world. They contribute to what the Editors call
‘the central question’: “How does the organization of complex syntax in real time
(i.e. in the very process of its production) relate to the on-line unfolding of turns
and actions, and hence to such fundamental tasks in social interaction as action
projection, formation and ascription?”
The attention to the unfolding of structure in real time differentiates the inter-
actional linguistics paradigm from structuralist approaches that present syntactic
form from a supervisory, ‘bird’s eye’ view in which forms exist simultaneously
from beginning to end. The temporality of spoken interaction is a theoretical key-
stone of interactional linguistics. Temporality, which is existential time, is to be
sharply distinguished from time. Hoy (2009: xiii) explains the difference in the
following way: “The term time can be used to refer to universal time, clock time,
or objective time. In contrast, temporality is time insofar as it manifests itself in
human existence.” Linguists’ account of time is generally limited to the classifica-
tion and description of structural details of the tenses and aspects coded by verbs,
adverbs and syntax.
Awareness of the way that linguistic form unfolds during acts of speaking is
fundamental to the procedures of interactive linguistics. As a matter of fact, the
relationship of temporality (as distinct from time) to speech, while it has only
recently become a central theme of linguistics, has not been ignored. The critic
and writer Gotthold Ephraim Lessing (1729–81) in his monograph Laokoon, oder,
Über die Grenzen der Malerey und Poesie (Laokoon or on the boundaries of painting
and poetry) (Lessing 1766) reasoned that there were two kinds of representation in
art: those that arranged their subjects in space, such as painting and sculpture, and
https://doi.org/10.1075/slsi.32.12hop
332 Paul J. Hopper
those that arranged their subjects in time, such as music and poetry. In German,
they were the Nebeneinander and the Nacheinander, juxtaposition and sequence.
Each kind had its own esthetic virtues, Lessing said. The novelist and playwright
Heinrich von Kleist (1777–1811) argued for a position that is in some respects
strikingly similar to that of interactional linguists. In his essay Über die allmähliche
Verfertigung der Gedanken beim Reden (The ongoing formulation of thoughts while
speaking) (Kleist 1806)1 he proposed that thought itself is generated through face-
to-face speaking encounters, and is structured step-by-step within speech as the
encounter continues:
The sequence of ideas and the expression of those ideas go forward in tandem,
and the mental processes of the two converge. Speech is not, therefore, a fetter, a
brake, as it were, on the wheel of the mind, but rather like a second wheel on the
same axle, running in parallel. (Kleist 1806) (tr PJH)2
Kleist and Lessing are only two of many thinkers over the ages who have been
preoccupied with the flux of time, often in the context of speech. The concern
of interactional linguists to study communication always within the constraints
of existential time brings the field of linguistics into step with a number of other
fields that during the latter 20th century oriented their thinking toward the reality
of the temporal environment. At this time a movement was afoot that rejected
the idea of a world of static substances that had dominated Western mentalities
through Plato and Descartes. In various guises the idea gained strength that reality
was processual and that essences were to be seen as distributed over time. One of
the most detailed formulations in recent times has been that of Nicholas Rescher,
whose book Process Metaphysics (Rescher 1996) together with the long article by
Johanna Seibt (Seibt 2012) concisely outline the arguments of process philosophy.
Seibt’s summary of this movement describes it as follows:
Process philosophy opposes ‘substance metaphysics,’ the dominant research
paradigm in the history of Western philosophy since Aristotle. Substance meta-
physics proceeds from the intuition – first formulated by the pre-Socratic Greek
philosopher Parmenides – that being should be thought of as simple, hence as
internally undifferentiated and unchangeable. Substance metaphysicians recast
this intuition as the claim that the primary units of reality (called “substances”)
1. For a fuller discussion of Kleist’s essay and its relevance to interactional linguistics, see Auer
2014.
2. Die Reihen der Vorstellungen und ihrer Bezeichnungen gehen neben einander fort, und
die Gemüthsacten für Eins und das Andere congruiren. Die Sprache ist alsdann keine Fessel,
etwa wie ein Hemmschuh an dem Rade des Geistes, sondern wie ein zweites mit ihm parallel
fortlaufendes Rad an seiner Axe.
Chapter 12. Afterword 333
must be static – they must be what they are at any instant in time. In contrast
to the substance-metaphysical snapshot view of reality, with its typical focus on
externalist being and on what there is, process philosophers analyze becoming
and what is occurring as well as ways of occurring (Seibt 2012).
The key priorities of process philosophy in contrast with substance philosophy as

laid out by Rescher (1996: 31) are as follows:
Table 1. Process vs. substance philosophy.

Process philosophy Substance philosophy
Activity Substance
Process Product
Change Persistence
Novelty Continuity
This formulation has obvious applications to language manifested as speech, al-

though strangely Rescher insofar as he discusses language at all holds to a standard
structuralist view, seeing speech itself as reflecting a fixed code that is, ironically,
not processual in nature. He marvels (Rescher 1996: 79) at the way a system with
fixed units like “letters, words, sentences, paragraphs …” can create an “infinite di-
versity” of structures, and elsewhere says that “…reality cannot be adequately (i.e.,
more than approximately) represented and conveyed by the relatively fixed and
discrete mechanisms of language and its concepts” (Rescher 1996: 135). Susanne
Langer, by contrast, recognized in her book Philosophy in a New Key (1947) the
special nature of the ‘discursive symbolism’ of language as different from the ‘pre-
sentational symbolism’ of other semiotic systems:
The meanings given through language are successively understood, and gathered
into a whole by the process called discourse; the meanings of all other symbolic
elements that compose a larger, articulate symbol are understood only through
the meaning of the whole, through their relations within the total structure.
Their very functioning as symbols depends on the fact that they are involved in
a simultaneous, integral presentation. This kind of semantic may be called “pre-
sentational symbolism,” to characterize its essential distinction from discursive
symbolism, or “language” proper. (Langer 1947: 78)
Whereas presentational symbolism offers a single symbol that is understood as a

whole, simultaneously, through its internal structure, discursive symbolism ap-
pears successively and achieves its meanings as the symbols are gathered together
in discourse. Langer’s discursive symbolism of meanings that are successively un-
derstood characterizes the approaches of all the papers in the current volume, and
of course the interactional linguistics paradigm in general as the editors describe
334 Paul J. Hopper
it in their Introduction. She places them in contrast to the structuralist standpoint

of other paradigms.
Dialogue is the second, though no less important, pillar of the interactional
paradigm. The cognitive role of dialogue was recognized by Heinrich von Kleist
in the same essay, with the adage l’idée vient en parlant (an idea comes in speak-
ing), a riff on the saying l’appétit vient en mangeant (appetite comes with eating).
Kleist tells how he struggles in vain to understand a legal document, or to solve a
mathematical equation:
Then when I talk about it to my sister as she sits behind me working, I learn things
that I could not have figured out in hours of brooding. It isn’t that she actually tells
me, for she has no knowledge of the law and has never studied Euler or Kästner.
Nor has she led me to the point at issue through astute questions …But because I
have an obscure idea that has a distant connection to what I’m looking for, if I just
make a bold start, letting my talk move forward with a view to carrying this begin-
ning through to its end, my mind etches the confused idea into complete clarity,
and so to my amazement enlightenment comes as my sentence finishes. I throw
in meaningless sounds, stretch out the conjunctions, perhaps use an apposition
where none was needed, and employ other devices in order to gain time for the
fabrication of my idea in the workshop of reason.3
With the observation that thought and argument require articulation in the
presence of another person, whether real or ‘virtual’ (see Linell 1998: 35), Kleist
approaches the assumptions of present-day interactional linguistics. According to
this perspective, utterances are constructed incrementally, in collaboration with
other speakers, who must constantly ratify them, and even contribute to their
construction. Structure emerges in step with the utterance, just as Kleist describes.
Grammatical form develops as it is assembled during acts of dialogic commu-
nication. Nowhere is this demonstrated more convincingly than in the on-line
3. Und siehe da, wenn ich mit meiner Schwester davon rede, welche hinter mir sitzt, und
arbeitet, so erfahre ich, was ich durch ein vielleicht stundenlanges Brüten nicht herausgebracht
haben würde. Nicht, als ob sie es mir im eigentlichen Sinne sagte; denn sie kennt weder das
Gesetzbuch, noch hat sie den Euler oder Kästner studirt. Auch nicht, als ob sie mich durch
geschickte Fragen auf den Punkt hinführte, auf welchen es ankommt, wenn schon dies letzte
häufig der Fall sein mag. Aber weil ich doch irgendeine dunkle Vorstellung habe, die mit dem,
was ich suche, von fern her in einiger Verbindung steht, so prägt, wenn ich nur dreist damit
den Anfang mache, das Gemüth, während die Rede fortschreitet, in der Nothwendigkeit, dem
Anfang nun auch ein Ende zu finden, jene verworrene Vorstellung zur völligen Deutlichkeit aus,
dergestalt, daß die Erkenntnis zu meinem Erstaunen mit der Periode fertig ist. Ich mische un-
artikulirte Töne ein, ziehe die Verbindungswörter in die Länge, gebrauche wol eine Apposition,
wo sie nicht nöthig wäre, und bediene mich anderer, die Rede ausdehnender Kunstgriffe, zur
Fabrikation meiner Idee auf der Werkstätte der Vernunft, die gehörige Zeit zu gewinnen.
creation of complex sentences through the incremental accumulation of clauses

and clausal fragments.
Structure then is not a solo performance but requires the participation of
another. The fragments of talk that are the material form of an interaction settle
into patterns and routines that reflect the organization of the interaction. As these
forms combine into longer arrangements, their structure becomes less fixed. The
interactants create ad hoc syntactic templates that rely on, but are not identical
to, previously used ones. These ad hoc templates are the source of the complex
sentences familiar to us from grammatical descriptions. They are “grammatical”,
but not in the standard sense of the word. “Grammar” in the standard sense is
a fixing or sedimentation. Standard theories of linguistics have looked down on
the fragmentary and “irregular” nature of ordinary conversation as “deviant”
and even “degenerate”:
Many children acquire first and second languages quite successfully even though
no special care is taken to teach it to them and no special attention is given to their
progress. It also seems apparent that much of the actual speech observed consists
of fragments and deviant expressions of a variety of sorts. (Chomsky 1965: 201)
The native speaker has acquired a grammar on the basis of very restricted and
degenerate evidence. (Chomsky 1972: 24)
However, when we consider the presumed perfection of whole sentences and

their privileged position in linguistic theory, as opposed to their rarity in natural
discourse, we must ask: Are fragments deviant forms of discourse, or is it rather
that the sentences of syntactic theoreticians are normativized and literate versions
of natural ways of talking, versions that have been elaborated in a centuries-old
intellectual and literary tradition?
There is an important conceptual difference between these two ways of
thinking about grammar. The standard idea of grammar is that it is something
static and complete. It is synchronic and therefore timeless. Emergent grammar
is temporal. Standard grammar is conceived as an abstract code that links a set
of spoken or written forms to a set of meanings and functions. Emergent gram-
mar gets its name and its concept from one overriding fact about speech, that
it takes place in material time and in material circumstances. This fact serves to
pace the delivery of a message because speakers are always responding to situa-
tions and especially to the other speakers, constantly gauging the likelihood that
they will understand and constantly trying to buy time while new discourse
situations are assessed and responded to. Emergent grammar is opposed to the
kind of pragmatics where “uses” like politeness and relevance are surmised on
the basis of imagined a priori situations, and to all introspective cognition, where
mental capacities are held to precede decisions about linguistic forms. The starting
336 Paul J. Hopper
assumption in these approaches is that a data set is surveyable simultaneously in

its opening, middle and concluding phases. An example of the difference is given
by Bakhtin in his essay The Problem of the Text. Noting that “[t]he understanding
of an utterance is always dialogic” (Bakhtin 1986: 125), Bakhtin goes on to say:
“One cannot, on the other hand, understand dialogic relations simplistically and
unilaterally, reducing them to contradiction, conflict, polemics, or disagreement.”
That is to say, ordinary utterances are always the result of a collaborative action
of a speaker and an interlocutor. We do not have two parties saying their pieces
separately. Rather the premises of each utterance are jointly arrived at by both par-
ties. It follows that the way we interpret utterances must always take into account
the existence of a support group who actively accept these premises; the support
“group”, of course, often consisting of simply a single listener. Suppose, and this is
Bakhtin’s own example, two people are talking and one says “Beautiful weather!”
and then the other also says “Beautiful weather!” (an exchange which is not so
remarkable in Russian as in English). One can imagine three ways or levels in
which this utterance pair could be interpreted. At one level the two utterances are
identical. This would be the logician’s view. It would not be possible for the two
utterances to have different truth values. At another level, the two utterances are in
some sense different, being spoken on different occasions, but they have enough in
common to warrant the second being understood as grammatically and lexically
the same as the first. The criteria for sameness and difference are the province of
the linguist or the grammarian; the search for such criteria has led to a refined
technical terminology of abstract structure, morphemes, and phonemes. Applying
the structural criteria of linguistics, we would say that the second utterance is a
repetition of the first. But there is a third level, that in which an act of agreement
has been performed. As Bakhtin says: “If they really are two utterances belonging
to different voices and not one, [they] are linked by a dialogic relation of agree-
ment. … For after all, agreement could also be lacking (‘No, not very nice weather,’
and so forth).” (Bakhtin 1986: 125) To understand, for Bakhtin, is not merely to
record and passively accept, but to react; in Bakhtin’s words:
[The speaker] does not expect passive understanding that, so to speak, only
duplicates his own idea in someone else’s mind. Rather, he expects response,
agreement, sympathy, objection, execution, and so forth.
The first two levels of interpretation have been called logical and linguistic or
grammatical. What of the third? It is rhetorical. It cannot be a coincidence that
these three domains of understanding correspond to the elementary core of the
medieval educational philosophy inherited from Aristotle, the Three-fold Way,
the Trivium: Logic, Grammar, Rhetoric, a grasp of which is so basic and so much
taken for granted that it has given us our modern word “trivial”. The term Rhetoric
is in a certain sense a paraphrase of the concept of emergent grammar. From the

emergent grammar perspective, grammar itself could be considered simply the
local detail of rhetoric, we might say micro-rhetoric (Hopper 2007). The province
of micro-rhetoric is the coordination of speech with other simultaneous activities.
Roy Harris’ “integrational linguistics” (see Wolf 1999) makes a similar argument,
criticizing the idea of a discrete language system segregated from the human
activities that accompany it. It would not make sense to study speech away from
these other activities, any more than it would make sense to study gestures without
some reference to the social activities with which they are integrated. Imagine, for
example, describing a shrug of the shoulders but making no mention of when such
a movement might occur, or even how, without a context, it could be distinguished
from a nervous twitch.
The papers in this volume represent the different ways in which linguists using
a variety of empirical data from a variety of languages work out the implications
of complex syntactic structure as it emerges from interactions. They represent a
radical departure from the attempt to derive the complex sentences of canonical
grammar from timeless uniform gestalts. Grammatical form moves forward in
time and is ever shifting, taking its cue from the speakers’ common experience of
the words of others. As Bakhtin put it, “I live in a world of others’ words” (Bakhtin
1986: 143). Speaking is a constant process of appropriating and adapting the words
of others. Hermann Paul observed that “…what was once understood or spoken
can be understood and spoken again” (Auer & Murray 2015: 49). Moreover,
nothing that is said escapes the mind of the hearer: “…not a single representation
formed in the mind through speech can be lost without a trace, even though this
trace will often be so weak that special circumstances …are necessary for it to be-
come conscious again” (Auer & Murray 2015: 49). That these traces are assembled
in real time in response to the needs called up by interactional situations is shown
again and again in the contributions to the volume.
References
Auer, P. (2014). ‘L’idée vient en parlant’: Kleists Entwurf zur dialogischen Emergenz von Sprache
und Denken. In W. Frick (Ed.), Heinrich von Kleist: Neue Ansichten eines rebellischen Klas-
sikers (pp. 63–85). Freiburg i.Br.: Rombach Verlag.
Auer, P., & Murray, R. W. (Eds.) (2015). Hermann Paul’s ‘Principles of language history’ revisited:
Translations and reflections. Berlin: De Gruyter. https://doi.org/10.1515/9783110348842
Bakhtin, M. M., (1986). The Problem of the Text. In C. Emerson, & M. Holquist (Eds.), Speech
genres and other late essays (pp. 103–131). Translated by Vern W. McGee. Austin: University
of Texas Press.
Chomsky, A. N. (1965). Aspects of the theory of syntax. Cambridge, MA: The MIT Press.
338 Paul J. Hopper
Chomsky, A. N. (1972). Language and mind. Cambridge: Cambridge University Press.

Hopper, P. J. (2007). Linguistics and Micro-Rhetoric: A Twenty-First Century Encounter. Jour-
nal of English Linguistics 35(3): 136–152. https://doi.org/10.1177/0075424207305307
Hoy, D. C. (2009). The time of our lives: A critical history of temporality. Cambridge, MA: The
MIT Press. https://doi.org/10.7551/mitpress/9780262013048.001.0001
Kleist, H. von (1806). Über die allmähliche Verfertigung der Gedanken beim Reden. http://guten-
berg.spiegel.de/buch/über-die-allmähliche-verfertigung-der-gedanken-beim-reden-589/1
(accessed Aug. 8 2018).
Langer, S. (1947). Philosophy in a new key. Cambridge, MA: Harvard University Press.
Lessing, G. E. (1766). Laokoon, oder, Über die Grenzen der Malerey und Poesie. http://gutenberg.
spiegel.de/buch/laokoon-1176/1 (accessed Aug 8 2018).
Linell, P. (1998). Approaching dialogue: Talk, interaction and contexts in dialogical perspectives.
Amsterdam/Philadelphia: John Benjamins. (Impact: Studies in Language and Society, 3)
https://doi.org/10.1075/impact.3
Rescher, N. (1996). Process metaphysics: An introduction to process philosophy. New York: SUNY
Press.
Seibt, J. (2012). Process philosophy. Stanford encyclopedia of philosophy. <http://plato.stanford.
edu/entries/process-philosophy > (Accessed July 2 2018)
Wolf, G. (1999). Editorial introduction. Integrational Linguistics in the Context of 20th-Century
Theories of Language. Special issue, Language and Communication 19(1), 1–8.
Index
A C complex syntax 1–4, 7, 10, 13,

account 5, 14, 61, 134, 146, clause combination 3, 12, 55, 58, 87–88, 118–119, 128,
245–249, 251, 253, 255, 257, 14–15, 31, 61–62, 66, 77, 154, 198, 303–305, 320–321, 324,
259, 261, 263–270 160, 173, 178, 245–246, 248, 331, 337
action combination 13–14, 250–251, 253, 258–260, 264, conditional clause 6–7, 187,
245–246, 250, 253, 258–259, 266–267, 269, 282n14 187n5, 213, 222, 251, 260,
267–269 clause-combining 1, 3, 7, 266
additive 161, 222 9–10, 12–13, 29, 55, 118–119, conjunctive phrase 27, 33, 50
conjunction 222 151, 153, 176, 178, 185–187, coordinating 201, 221–222,
adverbial 61, 87, 161, 253n5, 196, 204, 206, 214–215, 227, 224–225, 230, 232, 235, 237,
263 245, 255, 264–266, 304–305, 240, 248, 258, 307, 324
clause 7, 11–12, 61, 87–89, 321, 323–324, 331 conjunction 13, 221–224,
99, 104–106, 116, 118–119, co-construction 2, 99, 115, 230, 234, 239, 240
246, 265 185, 197–198, 200, 201n22, coordination 4, 12–13, 185,
complement 4, 104–105, 215–216, 284, 293, 297 221–223, 226–227, 234,
111, 117 cognitive verbs 55, 57, 76, 240–241, 246, 306–307, 337
conjunction 89, 104–106, 79–81, 151, 154, 161, 177 coordinator 13
118, 118n24, 119 complement 11, 55–58, 63, copula 6, 114, 128, 128n3, 129,
connective 251, 253 66–67, 70–74, 76–78, 78n14, 130–132, 135, 145
connector 260 79, 81, 87, 104, 105n18, 152, copular clauses 277n5
agreeing 130, 135–136, 154, 160, 163, 173–174, 178 co-reference 284, 290
145–146, 160, 174, 176, 209, clause 11–12, 14, 61–62, co-referential 95, 276–277,
315, 336 66, 87–89, 99, 116–119, 281, 284, 284n21
see also disagreeing 151–152, 154–156, 161, corrective actions 255
aligning 7, 99, 133, 136, 161, 173–174, 176, 178, 203, correferent 95
163, 169, 197, 199–200, 215, 251, 253, 275–277, 277n5, cosubordinate 88, 95, 119
307 279–280, 281n12, 282–283, see also insubordinate
see also disaligning 283n17, 284–286, 286n24,
assessment 29, 95, 98, 129, 287–289, 289n25, 291, 293, D
152, 161, 172n4, 175, 212, 295–298 demonstrative 14, 69, 78n14,
214–215, 275–276, 290–291, -taking 4, 12, 56, 56n3, 89, 95, 95n11, 109n19, 114,
297–298, 298n32, 307, 311, 80–81, 152, 154, 177, 282, 118, 161, 275–277, 277n2,
315, 335 297, 304 277n3, 278, 280n10,
see also adverbial 283–284, 284n21, 288, 290,
B complement 295, 296n30
bi-clausal 4–5, 13, 78, 89, complementation 55–56, deontic authority 257, 269
118, 129, 145, 186–187, 189, 56n3, 58, 66, 76 deontic rights 61
201, 215–216, 253, 264, 282, complementizer 11, 56, 58, dependency 4, 6, 13, 33,
298, 318 62–63, 66–67, 77, 79–80, 88–90, 95, 103, 103n17, 116,
construction 4, 9, 14, 117, 105n18, 118, 128, 128n4, 129, 119, 193, 196, 209, 214, 251,
118n23, 119, 129, 187, 189, 132, 145, 275–277, 280n8, 260, 268
275–276, 284n20, 298 288, 304 see also independency
bird’s eye view 1, 331
340 Emergent Syntax for Conversation
dialogic 186, 197–198, 206, English 2, 4, 4n2, 5, 7, 38n11, formulaic fragment 11, 67, 78,
249, 268–269, 334, 336 51, 56n1, 56n3, 58n8, 64, 81, 129
dialogue 15, 197, 334 64n10, 69, 74, 77, 95, 97, fragment 3–4, 6, 73–74, 80,
directive 14, 210, 210n34, 97n12, 98n13, 109n20, 127, 130, 145, 147, 186n2,
210n35, 229–230, 235, 237, 118n23, 129, 144, 151, 153, 187n5, 214n38, 251, 331, 335
245–253, 255–259, 261, 160, 177, 187n5, 210n34, frame shift 132
263–270 221–222, 234, 240, 266, French 2, 4–5, 7, 10, 15, 51,
disagreeing 42, 44–45, 51, 136, 280n11, 282, 283n17, 297, 56n3, 76, 76n12, 177n5, 231,
146, 199, 289, 323, 336 305–307, 336 282n13, 298, 298n33, 303,
disaligning 69, 144, 146 epistemic 11–12, 57n5, 61, 305–308
discourse marker 4–6, 12, 51, 69, 73–81, 146, 151–153, 155, frequency effects 153, 177
93, 101, 110, 119, 131–132, 136, 157–160, 163, 173, 177–178,
145–146, 177n5, 189 263, 282–283, 286–287, 297, G
311–312, 318 German 2, 4n2, 5, 10, 13–14,
E stance 4, 11, 59, 62, 66, 77, 56n3, 58n8, 64, 64n10, 77,
elaboration 11, 25, 31, 35, 80, 91n8, 151, 155, 157, 160, 99, 118n23, 129, 132, 145,
37n10, 39, 87, 90–91, 91n7, 173, 289 185–187, 188n7, 189n8,
99–101, 103–106, 110, Estonian 2, 4n2, 10, 13, 51, 190n10, 196, 196n15, 206,
112–117, 117n22, 118, 129, 146, 56n1, 76, 221–224, 230, 240, 210n35, 215, 234, 266–267,
250, 269, 318–321 282n14 275–277, 279, 280n11,
see also referential evaluation 11–12, 63–64, 282–283, 283n17, 297–298,
elaboration 87, 95, 98, 106, 116, 118, 332
embedded 4, 6, 49, 61, 87–88, 133, 135–136, 141, 145–146, gestalt 186–187, 193, 198–200,
95, 167, 169, 320 155–161, 163, 165, 172–178, 206, 214–215, 337
see also unembedded 253, 289 gesture 3, 9, 11, 44–45,
embodied 3, 9–11, 13, 25, evaluative 12, 63, 90–91, 91n7, 48–49, 98–99, 101–102, 106,
38–39, 44, 49–50, 87, 110, 96, 99, 116, 118, 151, 153, 109–110, 112, 116, 118, 239,
221, 223, 230, 235, 240–241, 155, 160, 163, 173, 175–178, 241, 259, 305, 337
245–250, 255–257, 259, 261, 282–283, 297, 311 grammar 1–4, 9–10, 12, 14,
264, 268–269, 303, 311, 320 stance 4, 11, 59, 62, 77, 80, 28, 89, 89n5, 117, 117n22,
behavior 13–14, 91n9, 97, 91, 95–96, 155, 160, 163, 127–129, 131, 147, 186, 213,
112, 221 172–173, 178 221–222, 237, 240–241,
conduct 1, 9–11, 26n2, 28, expansion 2–3, 8, 11, 13, 39, 246, 268, 276, 279, 283n17,
37, 43–44, 47, 117, 303–304, 151, 153–154, 163, 165, 167, 303–305, 308, 322, 324,
307–309, 311–314, 319, 324 172–174, 177–178, 185, 201, 335–337
emergence 14, 47, 49, 55–57, 203, 245, 248–249, 259, see also emergent grammar
87, 119, 127, 151, 174, 185–186, 264, 267, 269–270, 276–277, grammaticalization 214, 297
198, 200–201, 215, 245, 277n5, 280, 282, 284, 288, grammaticization 4, 12,
249–250, 257–259, 261, 264, 291, 293, 295, 298, 304 87–88, 118, 151–152, 178
268, 281n12, 304, 321
emergent 1, 7, 9–10, 12–13, 71, F H
127, 199, 255, 276–277, 293, Finnish 2, 4n2, 7, 10–11, 51, Hebrew 2, 4n2, 5–6, 10–12,
298, 303–305, 321, 324, 331 55, 56n1, 56n3, 56n4, 57, 74, 76, 87–89, 89n5, 90, 95,
grammar 7, 186, 245, 335, 337 57n5, 58n8, 59, 64, 67, 70, 95n11, 98n13, 99, 111, 114,
emerging 1, 3–4, 7, 10–11, 74, 77, 79–81, 187n5, 205n29, 115n21, 116, 117n22, 118–119,
13–14, 31, 76, 81, 99, 110, 210n34, 210n35, 213, 127–128, 128n2, 128n3,
127, 133, 143, 153–154, 165, 221–222, 259, 305, 307 128n4, 129–133, 145–146,
167, 177, 197, 201, 203–204, fixed 11, 62, 67, 81, 127, 130, 177n5, 307
206, 221, 223, 229–231, 146, 152, 160, 177, 215–216,
236–237, 239–241, 245–249, 333, 335 I
264, 268–270, 289, 291, see also semi-fixed increment 2–3, 6–7, 9, 14–15,
298, 303–304, 312, 318–319, fixedness 55–56, 67 29, 33, 38, 69, 95–96, 99,
321–323, 334, 337 103, 106, 110, 113, 116, 144,
Index 341
201, 203, 248, 259, 267, matrix clause 12, 14, 88, 89n5, projected 35, 39, 132, 167, 195,
275–277, 280n9, 289, 298, 95, 99, 104, 105n18, 114–116, 198–200, 245, 248, 266,
303–305, 307–309, 311, 119, 151–152, 160, 173–174, 269–270, 320
314–318, 320–323, 334–335 176, 178, 185, 206, 251, 253, see also non-projected
see also other-increment; 276–277, 277n3, 278–280, projecting 1–6, 9–12, 14, 31,
self-increment 281n12, 282, 282n16, 283, 33, 50–51, 55, 57, 66–67,
independency 5, 12–13, 283n17, 284–286, 286n24, 69, 77–78, 87, 99, 116, 118,
39, 80, 88, 129, 165, 167, 288–290, 293, 296, 296n29, 127, 130, 132, 135, 145–146,
170–173, 176–178, 191, 193, 296n30, 297–298 163, 165, 169–170, 172–173,
196, 206–208, 213–214, 240, matrix verb 117, 286 175–177, 177n5, 187, 189, 197,
263, 277, 279, 297, 304, 306, meaning 3, 12, 14, 32–33, 39, 203, 205, 213, 215, 226–227,
315, 323 50, 59, 64–66, 98n13, 154, 249, 257, 259, 262, 279–280,
instruction 14, 29–30, 195–196, 216, 222, 257, 275, 282n14, 283, 291, 320
225–227, 229–232, 234–235, 277, 282n16, 289, 297–298, construction 4–5, 10, 12, 127,
237, 239, 245–270 304, 333, 335 129–130, 132, 145, 147, 197
insubordinate 4, 6, 11, 13, micro-rhetoric 337 see also non-projecting
87–91, 91n7, 95–96, 99, modalities 14, 221, 240, 269 projection 2–4, 9–11, 13,
101n15, 103, 106, 110, 112, modality 214 57, 132, 167, 185–186, 203,
114–119, 185–186, 209, multimodal 3, 9, 88, 90, 116, 203n26, 205, 216, 245, 249,
209n33, 210, 212, 214, 323 128n2, 128n4, 178, 221, 223, 256, 264–265, 267–268, 270,
see also cosubordinate 239, 245–246, 250, 303–304, 276, 280–282, 296, 331
intersubjectivity 70, 154, 308, 311 projective 11–12, 31–32, 57,
160–163, 173, 177, 200, 215, multi-unit 163, 165, 167, 73, 79–81, 143–145, 177n5,
246, 250, 263, 315–316, 321, 169–170, 172, 201, 201n23 214–215, 223, 278
324 turn 11–12, 14, 35, 40, 167, projector construction 25, 31,
intonation 27, 31, 34–35, 37, 173, 215, 275, 278, 284, 291, 39, 50–51, 304
41, 43, 48–50, 94, 96–98, 297–298 projector phrase 10, 31, 42
98n13, 101–102, 103n17, 106, multi-word conjunction prosody 1, 3, 8, 10, 12–14,
112–116, 119, 132, 167, 171, 26–27 25, 31, 34–35, 35n9, 39–41,
173, 178, 189–190, 212–213, 49–50, 70, 70n11, 72–75,
223, 230, 240, 248–249, N 87, 95, 103n17, 105n18, 112,
259, 266, 269–270, 296, 310, N-be-that-constructions 116, 129, 132, 152–153, 167,
313–314, 316–318 129–130, 132, 197 169–173, 178, 186, 190–191,
contour 11, 34, 87, 93, 97, non-projected 295, 297, 321, 195–197, 206, 208–209,
98n13, 99, 101–103, 112, 323 212–214, 223–224, 227, 230,
116, 131–132, 134, 143, 145, non-projecting 206, 215 236, 240–241, 245, 248,
191, 193, 230, 235, 239–240, non-referential 280n10, 283, 250–251, 257, 259, 263–267,
266, 279–280, 280n11, 284n21, 289n26 269, 276, 278–281, 283, 286,
287, 289 non-responsive 74 289, 289n26, 296–297, 304,
unit 89n4, 90, 93–94, 307, 319–320, 323
94n10, 104, 105n18, 109, O prospective indexical 12, 153,
112, 131–132, 135, 144–145, other-increment 314, 323 172–173, 177
153, 163, 165, 167, 169–170, pseudo-cleft 5–6, 8, 14, 89,
174, 191, 193, 204–205, P 118n23, 197, 298
227, 227n1, 230, 240, 264, phonetic reduction 11, 55, 58,
278n7, 279–280, 280n8, 62–63, 66, 70, 72–73, 80–81 Q
280n11, 291, 298 pivot 8 question 5, 7, 11, 29, 31–32, 35,
Italian 2, 10, 25–27, 27n3, 28, constructions 9 40, 50, 69–70, 74, 77, 78n14,
41–42, 51, 99, 210n34 -like constructions 8 89, 104, 115, 144, 165, 205,
polarity 11, 55–59, 76–77, 80 222, 234, 267, 276, 284, 286,
M predicative clause 12, 127–128, 293, 295, 297–298, 298n32,
Mandarin 2, 10, 12, 56n1, 129, 130–131 313
151–153, 177–178 projectability 177–178
342 Emergent Syntax for Conversation
R repetition 31, 66, 113, 144, 158, 286, 298n33, 304, 307–309,
reduction 4, 11, 55, 58–59, 165, 175–176, 224, 227, 230, 322–323
61n9, 62–63, 66, 70, 72–73, 234, 249, 291, 293, 313–314, specification 11, 14, 39, 99,
79–81, 118n24, 132, 145, 152, 336 103–105, 117, 136, 163, 172,
170–171, 177, 227, 234 repetitive 222, 224, 227, 196n15, 275–277, 277n3, 279,
reference 14, 69, 160, 259, 230–231, 234, 240–241 280n10, 284–286, 286n24,
261, 275–277, 277n3, 277n4, response 1, 3–7, 11, 13–15, 29, 288–289, 293, 297–298,
278n6, 279, 284, 286–287, 31, 34, 38–39, 44, 46–47, 61, 311–312, 315
289, 289n26, 291, 293, 295, 65–66, 69–71, 73–74, 76–77, stance 4, 11–13, 57, 59, 62,
297–298, 303, 315, 319 93, 95, 97–98, 98n13, 99, 67, 74, 76–81, 91, 95, 98,
recognition 314 101–102, 104, 113, 115, 129, 127, 129–130, 133–136, 138,
see also co-reference 134–136, 144, 146, 159–160, 141, 145–147, 155, 157–158,
referent 118, 141, 275, 277, 163, 170, 172, 177, 187, 191, 160–163, 178, 199–200,
277n4, 283–285, 293, 206, 208–210, 216, 221, 212–215, 291, 307
297–298, 305, 307, 311, 223–224, 227n1, 229–230, see also epistemic stance;
314–318, 320–321 240, 248–249, 254–255, 258, evaluative stance
identification 314, 318, 320, 260–261, 263–264, 279, 287, stand-alone 114, 185–187,
322 293, 296, 303–305, 308–309, 187n5, 206, 207n30,
recognition 15, 303, 305, 311–313, 315–318, 320–322, 208–210, 210n34, 212–215
310–312, 316–318, 320–322 335–337 subordinate 4, 6, 11–15, 59,
see also correferent responsive 13–14, 28–29, 31, 61, 66, 78, 87–89, 89n5,
referential 116, 161, 276, 49–50, 55, 57–58, 67, 70–71, 90, 116, 118–119, 128, 145,
280n10, 284, 305, 312, 74, 76–77, 79, 81, 110, 144, 147, 185–186, 189, 191, 193,
315–317, 319, 321–323 163, 239, 241, 286 195–196, 205–210, 213–214,
elaboration 15, 303, 305, 309, see also non-responsive 248, 251, 253–254, 260, 263,
316, 320 rhetoric 336–337 275, 303, 305–306, 323
identification 315, 322 see also micro-rhetoric see also cosubordinate;
problem 311, 313–314 rhetorical 129, 205, 212, 277, insubordinate
repair 15, 282n13, 293, 295, 288, 336 subordinator 6, 89, 89n5,
303, 305, 307, 309, 312, right-dislocation 14, 275–279, 253n5, 263
314–316, 321–323 281, 281n12, 282, 282n13, Swedish 2, 7, 9–10, 71, 99,
see also co-referential; 282n16, 283, 283n17, 284, 205n29, 210n34, 213, 234,
non-referential 284n20, 286, 288–289, 291, 240, 246, 250–253, 253n5,
referring 95, 141n6, 145, 160, 293, 297–298, 316 267, 270
163, 275, 277, 279, 284, routinization 1, 3–6, 10–12, synchronization 110, 117,
284n22, 284n23, 285–289, 55, 79, 117, 127, 185, 206, 210, 197–198, 221, 225, 227,
289n25, 290, 290n27, 293, 214–215, 315 230–231, 234–235, 237, 240
295–296 syntactically integrated 11–13,
relative 25, 87, 104, 305–308, S 87–88, 90, 109, 117–118,
323 sedimentation 1, 10, 12–13, 118n24, 119, 185, 187, 191, 195,
clause 7–8, 11–12, 14–15, 185–186, 191, 201, 335 200n21, 206, 215
25–27, 39, 50, 87–89, 89n5, self-increment 15, 303–305, syntactically loosely-
95, 99, 103, 103n17, 109, 309, 312, 314–316, 318, integrated 13, 89–91,
109n19, 111, 115, 115n21, 320–321 103–104, 109n19, 111, 116,
116–117, 117n22, 118–119, self-repair 40–41, 50–51, 104, 118–119, 185
268, 303–307, 307n2, 110, 112–113, 293, 297, 315 syntactically unintegrated 11,
308, 321 semi-fixed 56, 76, 79–80 13, 87–91, 114–116, 118–119,
relativizer 26, 89, 118 sequential 2, 6, 10, 13, 31, 185, 191–192, 195, 197,
repair 29–31, 248, 281, 293, 49, 55–58, 67, 71, 76–77, 200n21, 207, 215
295, 297, 303, 305, 311–312, 81, 90, 127, 130, 144, 146,
314–316, 318, 320–323 153, 178, 197, 201, 204, 224, T
see also referential repair; 245, 249–250, 252n4, 255, TCU 9, 13, 25, 27, 29–31,
self-repair 257, 259, 261, 264, 283–284, 33–36, 40–41, 49–51, 57, 67,
Index 343
71, 77, 131, 193, 201, 204n27, trouble 7–8, 15, 303, 305, U
205, 215, 284, 284n23, 288, 310–313, 315, 317–318, understanding 25, 31–34,
290, 290n27, 295, 298, 320–322 38–40, 50–51, 61, 69, 95,
308–309, 323 turn-construction 28, 33, 37, 160, 241, 249–250, 257,
temporality 15, 89, 222–223, 40, 49–50, 282n13 263–264, 268–269, 277, 279,
249, 298, 331 turn-expansion 39, 151, 286, 293, 296, 296n29, 298,
time 1–3, 7, 9, 12–13, 15, 186, 153–154, 163, 165, 167, 315, 321, 333–337
197–198, 203, 203n26, 172–174, 177–178, 288 unembedded 95, 99, 116, 119
204–206, 222–223, 226–227, turn-initial 27, 29, 57, 76,
229, 231, 236, 239, 241, 76n12, 77, 91, 112, 114, 116,
303–304, 321, 323, 331–335, 131, 144, 146, 152, 163, 172,
337 284
trivium 336 turn-transition 31, 34, 39,
48–49
This volume explores how emergent patterns of complex syntax – that
is, syntactic structures beyond a simple clause – relate to the local
contingencies of action formation in social interaction. It examines
both the on-line emergence of clause-combining patterns as they
are ‘patched together’ on the fly, as well as their routinization and
sedimentation into new grammatical patterns across a range of
languages – English, Estonian, Finnish, French, German, Hebrew,
Italian, Mandarin, and Swedish.
The chapters investigate how the real-time organization of
complex syntax relates to the unfolding of turns and actions,
focusing on: (i) how complex syntactic patterns, or routinized
fragments of ‘canonical’ patterns, serve as resources for projection,
(ii) how complex syntactic patterns emerge incrementally, moment-
by-moment, out of the real-time trajectories of action, (iii) how formal
variants of such patterns relate to social action, and (iv) how all of
these play out within the multimodal ecologies of action formation.
The empirical findings presented in this volume lend support to a
conception of syntax as fundamentally temporal, emergent, dialogic,
sensitive to local interactional contingencies, and interwoven with
other semiotic resources.
isbn 978 90 272 0443 1

Emergent Syntax For Conversation - Clausal Patterns and The Organization of Action

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

Emergent Syntax For Conversation - Clausal Patterns and The Organization of Action

Uploaded by

Copyright:

Available Formats

Emergent Syntax

John Benjamins Publishing Company

Simona Pekarek Doehler

John Benjamins Publishing Company

the American National Standard for Information Sciences – Permanence

© 2020 – John Benjamins B.V.

Part I. Emerging projecting constructions

Part II. Locally emergent clause-combining patterns

Simona Pekarek Doehler, Yael Maschler, Leelo Keevallik and

or routinized fragments of what are considered ‘canonical’ patterns, serve as

As a consequence of this new conceptualization of grammar as a locally sensi-

Pekarek Doehler, 2011a, 2015)1 or practices for projection of bodily demonstra-

(1) (Pekarek Doehler, 2016, Coll_JSP_90)

02 (0.8) ((sliding sound, followed by two short knocks))

03 JUL: be:n >ch’p↑as< je trouve c’est mi↑eux^he.

105 ...hi lo tsrixa lehagid l-o,

In this Swedish conversation, the conditional clause in line 3 can be argued to

11 Vikke: just vuaden?, se, Mäkelän Kimmon kaveri.

This turn is composed of two syntactically incomplete clauses, each one

3. Purpose and structure of the volume

i. Emerging projecting constructions

conversation. The phrase, which according to traditional understanding projects

integrated varieties of Hebrew relative, complement, and adverbial clauses stand

ii. Locally emergent clause-combining patterns

understanding of what kinds of interactional tasks such on-line grammatical pat-

in French talk-in-interaction. It argues that a clausal combination consisting of a

Auer, P. (1992). The neverending sentence: Rightward expansion in spoken language. In M.

Pekarek Doehler, S. (2011a). Clause-combining and the sequencing of actions: Projector

Emerging projecting constructions

Nel senso (che) in Italian conversation

Keywords: turn position, projector construction, Italian, syntax

che is a multi-word conjunction that enables Giulia to extend a possibly complete

These illustrations show regular manifestations of che in Italian, which may be

monitoring” a preceding portion of talk; turn-final uses appear to achieve “boost-

3. Data and method

The analysis is based on a collection of 19 instances of nel senso and 25 instances of

as TCUs (e.g., Couper-Kuhlen, 2012; Lindström, 2006; Selting, 2005), especially

4. Turn-initial nel senso (che) in responsive actions

19 mettere le ruote s[enza che tu metta^in&

14 giugno weekend venerdì domenica. gli mettiamo le

(5) Ca13CL2–10 (20: 12–20: 28)

the teacher makes relevant. An orientation towards a potential problem of mean-

5. Nel senso (che) and turn-construction

On occasion, speakers may extend a turn beyond a place of possible completion by

5.1 Turn-extensions after other-speaker’s talk

Schegloff (2016, p. 253) observed that increments may occur “post-other-talk.”

5.2.1 Elaborating on a previous action

clause 1: dalle sollecitazioni offerte^è venuto fuori:

07 un for*te* .h (.) pensiero (.) a(l) priori.*

We observe again similar turn-constructional features as in Excerpt (7): Pamela

Figure 1. Paolo (and other participants) looking at Pamela

Figure 2. Paolo vigorously nodding Figure 3. Paolo vigorously nodding

08 tras- stra- strategia di sviluppo.

5.2.2 Initiating a new action

02 marketing. (0.4) di questa parleremo domani. (.)

(12) Mi13CE1–39 (07: 21–07: 35)

of his statement as a complete clause (per venticinque ti offro anche un serv(h)izio

Figure 4. Rino looking at his hand

Figure 5. Rino looking at Angela

By gazing at Angela, Rino addresses her as an appropriate recipient of his talk

Figure 6. Rino moving left hand outwards

Figure 7. Rino gazes at Giorgio, hand resting

At lines 03–06, Piera relays what is recognizable as a personal experience of

07 un forte .h (.) pensiero (.) a(l) priori.*