New Ideas in Psychology: Gregory J. Mills

New Ideas in Psychology 32 (2014) 158173
Contents lists available at SciVerse ScienceDirect
New Ideas in Psychology

journal homepage: www.elsevier.com/locate/newideapsych
Dialogue in joint activity: Complementarity, convergence

and conventionalization
Gregory J. Mills
University of Edinburgh, United Kingdom
a b s t r a c t
Keywords:
Dialogue
Grounding
Alignment
(Mis)communication
Coordination
Sequentiality
Dialogue is tightly interwoven within everyday joint activities that require moment-bymoment coordination of utterances and actions. A common account of coordination is
that it is established via progressive convergence (alignment, entrainment, similarity) of
interlocutors representations and behaviour. In order to examine how coordination is
established and sustained, this paper distinguishes between (1) Semantic coordination of
referring expressions (2) Procedural coordination of the timing and sequencing of contributions. Drawing on data from a series of maze experiments, this paper shows how both
kinds of coordination result in the rapid development of highly elliptical, systematized and
normative conventions. Focussing on how these conventions are established, this paper
shows how interlocutors exploit partial repetition as an interactive resource, resulting in
interlocutors turns becoming progressively divergent and complementary. Further, this
paper develops the claim that since repetition is best conceived as a special case of
complementarity, it cannot be the general explanation of coordination.
2013 Elsevier Ltd. All rights reserved.
1. Introduction
furniture up a ight of stairs must communicate momentby-moment in order to establish when and how to lift.
Finally, even in the absence of any overt physical actions,
dialogue is sui generis analyzable as a joint activity (Clark,
1996). Interlocutors must collaboratively negotiate how to
transition through different stages in the conversation; the
form of this negotiation depends strongly on the type of
conversation (e.g. story-telling, gossiping, enquiring about
a products price, or inviting friends for dinner) and here
too, successful coordination can also require the use of
activity-specic expressions and routines.
These insights have yielded theoretical units of analysis
that take into account the relationships between multiple
utterances and actions, e.g. language game (Wittgenstein,
1958), speech genre (Bakhtin, 1986), activity type
(Levinson, 1992), speech act1 (Austin, 1962), adjacency pair
(Schegloff, 2007), joint project (Clark, 1996), scripts (Schank
& Abelson, 1977), communicative project (Linell, 1998).
Dialogue in joint activities is the most basic form of

language use. We learn language via dialogical interaction,
and it is in our everyday conversational interactions that
we use language. A central feature of dialogue is that it is
intertwined with our activities in a myriad ways: for
example, conversing with a friend while walking down the
street, joking while passing food around the table, or
making small-talk while buying something at a shop. Dialogue can also serve to elaborate or augment an ongoing
activity, for example a dentist explaining the stages of a
procedure, whether for the comfort of a patient or to
instruct a student. Often, successful performance of an activity depends intrinsically on dialogue, in particular on
activity-specic utterances that coordinate how the activity
unfolds. Purchasing an item in a shop requires performing
the correct requests, actions and responses in the correct
order. Similarly, two people manoeuvering a large piece of
E-mail address: [email protected].
0732-118X/$ see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.newideapsych.2013.03.006
1
This only applies to Austins speech acts, as Searles subsequent
formalization (1969) removed the requirement of uptake by the hearer.
G.J. Mills / New Ideas in Psychology 32 (2014) 158173
Based on these insights, a vast body of work has uncovered

the rich variety of communicative devices (i.e. conventionalized, interactive routines) used by interlocutors to
both establish and sustain coordination in joint activities,
e.g. the use of gestures or body-posture to signal readiness
to engage (Kendon, 1976; Schegloff, 1998), the use of special
kinds of utterance such as outlouds (Heath & Luff, 1992)
or other signals to suspend and resume the activity
(Bangerter & Clark, 2003), as well as the use of physical
artefacts to coordinate the activity (Hutchins, 1995;
Scribner, 1986).
Using these kinds of routines seamlessly in the interaction can require extremely high levels of moment-bymoment phonological, lexical, syntactic and semantic coordination of both utterances and actions, between multiple interlocutors. This can involve knowing specialized
vocabularies, knowing which behaviour is sanctioned,
which complementary actions, utterances and associated
roles your interlocutors are performing, which roles and
associated actions you are expected to perform, and when
and how to perform them. Even in non-verbal activities,
studies of joint action have demonstrated the rapidity with
which participants establish coordination that allows them
to predict the timing, spatial orientation and format of each
others actions (Sebanz, Bekkering, & Knoblich, 2006).
Despite its centrality for any theory of dialogue there
has been a paucity of experimental studies that systematically address how interactive routines become conventionalized in the rst place. On the one hand, studies of
interactive practices (e.g. situated cognition, ethnographic
studies, conversation analysis) typically restrict their analyses to single (or few) episodes of interaction. The analysts
task is to uncover the (often highly tacit and activityspecic) conventionalized routines used by interlocutors,
yielding highly detailed analyses of short stretches of talk.
However, these studies do not systematically address how
two or more interlocutors, when encountering a novel
situation, interactively conventionalize novel routines for
coordinating with each other in the activity ab initio. Nor do
these studies systematically address how these conventions might be transformed on each occasion of their use
(whether successful or unsuccessful), as interlocutors
become progressively coordinated.
On the other hand, experimental approaches that do
study the emergence of conventions in dialogue typically
restrict their analyses to the study of referring conventions,
also eschewing systematic analysis of how the interactive
routines that yield these referring conventions are established and sustained. Within these approaches, conventionalization in dialogue is typically framed as a problem of
how two (or more) participants converge on the same
referring expressions. The immediate question that emerges
is: How amenable are the basic coordination mechanisms
proposed by existing accounts of referential conventions for
explaining how interactive routines become established?
Can the development of routines be adequately accounted
for as a form of progressive convergence?
The point of departure of this paper is to clarify how
existing models account for convergence. First, this paper
argues that convergence presents interlocutors with a SEMANTIC COORDINATION problem: interlocutors must adapt the
159
semantics of their referring expression to each other and

the demands of the activity. The development of semantic
coordination exhibits patterns of repetition that cannot be
adequately captured with existing models. Second, coordinating on routines requires PROCEDURAL COORDINATION of the
timing and sequencing of contributions. Procedural coordination is underpinned by interlocutors making COMPLEMENTARY and different contributions. The progressivity (i.e.
forward momentum) (Schegloff, 2007) of an interaction
depends intrinsically on interlocutors not repeating each
others utterances. Since repetition in dialogue is a special
case of complementarity, it cannot be the general mechanism behind coordination.
To examine separately how both kinds of coordination
are established and sustained, this paper draws on ndings from a collaborative maze task. In accounting for the
development of SEMANTIC COORDINATION, this paper draws
attention to the central role played by partial repetition.
The proposal advanced here is that turn-by-turn repetition
of structure is best conceived as a form of scaffolding
which supports the repair of existing representations and
also supports the construction of novel representations.
While some of these supporting structures might become
integrated into the resulting representation, they need not
be; some may be reused to construct other representations, and others may simply be used on a single occasion
and then discarded. Convergence does not arise straightforwardly out of repetition the structures that are
repeated turn-by-turn are not the same structures that
end up being converged on over the course of the
interaction.
In accounting for the development of PROCEDURAL COORDINATION, this paper argues that from the outset of the interaction participants are orientated towards the
COMPLEMENTARY structure of their contributions. Interlocutors
use communicative devices for establishing complementary structures, in particular for anchoring contributions at
sequential and temporal junctures within the activity. As
procedural coordination develops, and the forward momentum of interlocutors contributions increases, interlocutors produce progressively divergent turns,
demonstrating that coordination cannot be explained as a
straightforward form of convergence.
On encountering a novel activity with a novel partner,
neither SEMANTIC nor PROCEDURAL coordination can be presupposed.However, onencounteringand resolvingbothkinds
of coordination problem, the solutions become progressively
rened and systematized. As coordination progresses, and the
activity becomes sufciently well-dened, this paper argues
that in addition to developing normative referring conventions (see Brennan & Clark, 1996, conceptual pacts), interlocutors also rapidly develop normative procedural
conventions (i.e. procedural pacts) for resolving the procedural coordination problems encountered in the activity.
2. Accounting for coordination in dialogue
2.1. Semantic coordination of referring expressions
One of the most basic ndings in studies on referring
conventions is that convergence requires interactive
160
negotiation between participants (Clark & Schaefer, 1989;

Krauss & Weinheimer, 1966). Crucially, the processes
involved in establishing an initial communication system
apply both to situations where no pre-existing conventions
exist (Galantucci, 2005; Garrod, Fay, Lee, Oberlander, &
MacLeod, 2007; Healey, Swobod, Umata, & King, 2007),
and also apply to situations for which participants already
possess an ontology and referring schema (Garrod &
Anderson, 1987). Even when referring expressions are
given experimentally, as in the map task (Anderson et al.,
1991), interlocutors still conventionalize novel ad-hoc semantic forms (Larsson, 2007). This suggests that successful
communication requires more than just individual interlocutors being able to individuate and name referents
the demands of the interaction place constraints on mutual
intelligibility which require interlocutors to interactively
adapt their ontologies and the semantics of their expressions to each other and to the particular context. Moreover,
if participants interact within a collaborative activity, on
repeated use, the semantics of the referring expressions
become more concise, stable, and systematic (Clark &
Schaefer, 1989; Fusaroli et al., 2012; Garrod & Doherty,
1994; Healey, 1997; Mills & Healey, 2006; Schwartz,
1995). As a result, the semantics of the referring conventions established at the end of the interaction at high levels
of coordination can differ qualitatively from those used
initially.
These ndings suggest that explaining the aetilogy of
conventions requires paying close attention to the development of coordination, in particular how global convergence that occurs over the course of an interaction arises
out of local turn-by-turn processes.
2.1.1. Is global convergence driven by local repetition?
Perhaps the most seductively straightforward account
of global convergence is that it is achieved via interlocutors
representations becoming progressively similar during the
interaction: as interlocutors are exposed to each others
referring expressions, they gradually repeat more and more
of each others contributions, yielding progressively
convergent referring expressions. All things being equal,
the more similar (converged, aligned, entrained) the representations, the more coordinated the joint activity.
This picture of dialogue nds its strongest expression in
the Interactive Alignment model of Pickering and Garrod
(2004), which proposes that highly coordinated dialogue
arises directly as a consequence of local turn-by-turn
repetition at every level of linguistic processing (phonetic,
lexical, syntactic, semantic, situational) and is further
shored up by entrainment of speech-accompanying gestures, body sway, eye gaze and other forms of idiosyncratic
behaviour. For each level of representation and also for
each link between levels, local turn-by-turn repetition
leads cumulatively (via priming) to global convergence
(alignment) in a winner-takes-all fashion, operating both
within- and between-speakers.
But can this picture of local and global repetition
adequately account for coordination? Aside from the
observation that, in extremis, repetition, e.g. of words
(Simon, 1975) or of roles (Conchiglia, Rocco, & Grossi, 2007)
is pathological, the local and global patterns observed in
task-oriented dialogue work against this straightforward

view:
First, accounts of global convergence arising out of local
repetition are inherently conservative (Healey, 2004).
Repetition as a mechanism only operates on pre-existing
representations, and is consequently ill-suited for
capturing novelty, except as a form of noise.
Secondly, such conservative accounts are ill-suited for
capturing how a convention, when used successfully, might
be supplanted by another (Healey, 1997). Yet, interlocutors
do not settle on the initially most frequently used
convention; they continue to develop novel, more systematic conventions throughout the interaction (Garrod &
Doherty, 1994) (see Table 1, below). Note that, although
the collaborative model of Clark (1996) does not argue that
convergence arises out of local repetition, it too is conservative, in that it provides no mechanisms for a successfully
used convention to be supplanted by another (Healey,
1997).
Third, existing accounts are semantically neutral,
rendering them insensitive to the development of systematicity (Healey, 2004; Mills & Healey, 2008). The
Interactive Alignment model, the Collaborative model as
well as Conversation Analytic approaches are agnostic
about the semantics of referring expressions, yet it matters
which words are repeated: indiscriminate alignment is
correlated with unsuccessful dialogue (Fusaroli et al., 2012).
Fourth, accounts of global convergence arising out of
local repetition typically rely on this being an automatic,
rapid process. However, repetition of anothers turn in
dialogue can have a multiplicity of strategic functions, e.g.
conrming allusions (Heritage & Raymond, 2005;
Schegloff, 1996), performing clarication (Schegloff, 1992)
or correction (Jefferson, 1987; Saxton, 1997), or closing a
sub-sequence of an activity (Schegloff, 2011). Even echolalia has been re-analysed as having a complex interactional function (Sterponi & Fasulo, 2010).
Fifth, Garrod (1999) raises the possibility that repetition
of anothers referring expression can have a different
function at different stages of the interaction, depending on
whether participants are more or less coordinated. Supporting this view, Mills (2007) found that on encountering
problematic reference, interlocutors repeated each others
semantic choices more, not less. See also Louwerse, Dale,
Bard, and Jeuniaux (2012), who propose that interlocutors
exploit synchrony as an interactive recovery device.
Table 1
Global development of coordination in the maze game.
Initially:
5 min:
10 min:
15 min:
20
30
35
40
45
min:
min:
min:
min:
min:
OK, my grey switchpoint which you have to get

to now is next to the top sticking out cluster
Thats me done, can you go two down from the large
block of squares on the right
Tell me yours and when I say, go onto the middle
row, 3rd square from the left
Now to open it, wait 3 seconds then go 3rd row,
2nd sq from left"
Wait then go 5th column top square
Went back to 5th column top
Now on 5 col, row 7 from left
Then 5c r 7. yours?
5, 7
Taken together, these ndings suggest that local repetition cannot be the primary mechanism giving rise to
progressive global convergence.
2.2. Procedural coordination of routines
Although existing accounts of dialogue differ in their
characterization of how coordination is achieved, they
agree that convergence depends on referring expressions
being embedded within a collaborative, meaningful joint
activity (Brennan, Galati, & Kuhlen, 2010; Brown-Schmidt &
Tanenhaus, 2008). Interfering with the activity, e.g. by
limiting interlocutors ability to provide each other with
feedback, impedes coordination: core dialogue phenomena
such as referential contraction (Clark, 1996; Krauss &
Weinheimer, 1966), alignment (Pickering & Garrod, 2004)
and audience design (Brennan et al., 2010; Gann & Barr,
2012) only occur as a consequence of highly tacit reciprocal adjustment by interlocutors to each others informational needs within the activity, as the interaction unfolds.
This raises an additional problem: In order for referring
expressions to be meaningful, they have to be associated
with referents that have a meaningful role in the activity,
and in order for these referents to have a meaningful role,
the referents must be used by interlocutors in ways that
have real, interactional consequences for their conversational partners, in particular for how the activity unfolds.
For convergence to occur, referring expressions must have
more than a simple referential function; they must be
embedded within routines that are consequential for
coordinating how the activity unfolds.
However, when encountering a novel partner, in a novel
activity, it cannot be assumed that interlocutors are already
fully coordinated on the interactive routines associated
with that activity. This warrants the question: what is the
basic structure of routines, and how are they established?
2.2.1. A more nuanced view: repetition as a special case of
complementarity
One of the fundamental insights of Conversation Analysis is that coordination in dialogue is underpinned by
sequential structures which consist of pairs of complementary contributions (Schegloff, 2007). These adjacency
pairs consist of a rst and second part that operate
normatively: production of the rst part creates an
expectation that the second half is accountably due
(Heritage, 1984), leading any response to be interpreted as
pertaining to the second half. This locally managed system
of local sequential coherence between turns results in
global coherence through the hierarchical interleaving of
embedded sequences that resolve local problems through,
e.g. clarication, elaboration and reformulation (Levinson,
1983).
Although some adjacency pairs involve local repetition
of anothers utterance, e.g. in greetings (Hi/Hi), or in
exiting from telephone conversations (Schegloff & Sacks,
1973), repetition in dialogue is a special case of complementarity. This stands to reason, dialogue is intrinsically
progressive (Schegloff, 2007) and enchronic (Eneld &
Sidnell, 2013). We dont pathologically repeat each others
utterances; in order for a conversation to have forward
161
momentum, interlocutors must necessarily do something

different than their interlocutor. Canonical examples of
adjacency pairs include: question/answer; praise/selfdeprecation; offer/acceptance (Levinson, 1983). Progressivity is not merely a structural issue but is a strong social
norm: in order to maintain the forward momentum of a
conversation, interlocutors will violate other norms (e.g.
turn-taking and selection of the next speaker) (Schegloff,
2007; Stivers & Robinson, 2006).
Complementarity is also basic to joint activities. Successfully coordinating within a joint activity requires
knowing the next relevant contribution (Clark, 1996:
p252); Non-verbal joint actions also typically consist of
complementary structures (e.g. giving vs. taking)
(Newman-Norlund, van Schie, van Zuijlen, & Bekkering,
2007) Importantly, both the rst and second pair parts of
complementary structures can be either an action or an
utterance (Clark & Krych, 2004). Further, individual pair
parts are typically associated with highly differentiated,
complementary roles, whether brief and alternating
(speaker vs. hearer) or more extended (customer vs.
salesperson).
But how do interlocutors know what the next relevant
contribution is, and when and how to perform it? This
raises a basic problem for both cognitive and conversation
analytic studies, as both presuppose that adjacency pairs
are already shared (and known to be shared) by
interlocutors.
2.2.2. Conventionalizing complementary structures
When engaging in a novel activity with novel interlocutors, it cannot be assumed that interlocutors already
know the rst or second pair part, or even the minimal joint
action that comprises both. However conversation analytic
and cognitive approaches to dialogue have eschewed systematic analysis of how adjacency pairs are established in
the rst place. Further, once an adjacency pair is established, neither approach addresses how the progressivity of
the same adjacency pair might change on successive use, or
how an adjacency pair might acquire another conventionalized second pair part (Schegloff, 2007: p251).
In addition, Schegloff (2007: p223) remarks that
different activities can have very different associated routines for coordinating the activity which must be deployed
at specic conventionalized junctures within the activity.
How are junctures established interactively?
Adjacency pars are also normative conventions. If at a
particular juncture an interlocutor cannot provide a second
pair part that complements the others action, interlocutors
typically give an account for their inability (Schegloff,
2007). Examples include: giving reasons for declining an
invitation or demonstrating inability to shake hands by
showing how both hands are carrying heavy items. How do
regularities of structural association between two pair
parts acquire normative status?
Solving these procedural coordination problems is in
the rst instance an interactive problem faced by interlocutors. Modelling these problems and their solutions
requires giving an account of how two different contributions come to be associated as part of a whole adjacency pair. Further, since one of the hallmarks of
162
coordinated dialogue is its progressivity, the development of procedural coordination necessarily involves the
differentiation of interlocutors turns as coordination increases. (see Mills, 2011, for an initial attempt to investigate procedural coordination separately from semantic
coordination).
Dyads must solve 12 randomly generated mazes (see

Fig. 1), permitting systematic investigation of how the
communication system develops on each iteration2 of the
joint activity.
2.3. How do semantic and procedural coordination develop?
Pairs of native English speaking university students

participated for credit. Participants were given up to 90 min
to complete 12 mazes. Dyads consisted of a mixture of
familiar and unfamiliar pairs.3
Participants sit in different rooms, in front of a
computer screen which displays two applications (1) The
maze application and (2) A text-based experimental chat
tool for communicating. The maze application displays a
simple maze consisting of a conguration of nodes that
are connected by paths to form grid-like mazes. The
mazes are selected so that participants encounter an
equal number of grid-like and irregular shaped mazes.
The black circle shows the players current position and
the cross represents the goal point that the player has to
reach. Participants can move their location markers from
one node to another via the paths. Each move is recorded and time-stamped by the server. The game requires
both participants to move their location markers from a
starting location to a goal. Although the basic maze topology is the same for both participants, each participant
has a different starting location, a different goal location,
different switches and different gates, none of which are
visible to the other participant. They are also not able to
see each others location markers. Movement through
the maze is impeded by gates that block some of the
paths between nodes. These gates can be opened by the
use of switches (grey coloured nodes). The locations of
switches and gates are different on each maze and are
not visible to the other participant. Whenever a participant moves to a node that is marked as a switch on the
others screen, all of the other participants gates open.
All the gates subsequently close when they move off the
switch.
These constraints force participants to collaborate: In
order for participantA to open their gates, A must guide B
onto a node that corresponds to a switch that is only visible
on As screen (and vice versa). Solving the mazes, therefore
(i.e. when both participants are on their respective goals)
requires participants to develop coherent ways of referring
to maze locations, while also requiring participants to
develop procedures for requesting, describing and
traversing switches, gates and goals.
All participants communication occurs via a text-based
experimental chat-tool (Mills & Healey, 2006, in
preparation). The chat-tool interface is equivalent to
existing proprietary chat-tools (Skype, Facebook).
When interlocutors establish a novel communication

system for a novel activity with a novel partner, neither
referential nor procedural coordination can be presupposed. This leads to the immediate question how are
both types of coordination established?
There is, prima facie, a tension between accounts of
conventionalization which conceive of coordination as
consisting primarily of the repetition of referring expressions (or, for that matter, of the turn by turn repetition of
any communicative behaviour), and the observation that
procedural coordination is underpinned by different and
complementary structures between interlocutors. Accounts
of coordination must explain how the development of coordination involves both systematic convergence of participants onto single referring expressions, while at the same
time involving systematic local divergence of participants
turns into complementary contributions. Put simply,
repeating anothers turn is a special, not general, kind of
next relevant contribution, so cannot be the general
solution.
This matter is made all the more pressing, as one of the
main families of measures used to investigate coordination
involves examining how repetition of structure (lexical,
syntactic, semantic) relates to the quality of the interaction,
in particular how repetition correlates with task success
(Ferreira & Bock, 2006; Fusaroli et al. 2012; Healey, Purver,
& Howes, 2010; Reitter, Moore, & Keller, 2006). To this end,
this paper demonstrates separately how semantic and
procedural coordination develop during a simple joint
activity.
3. Method
3.1. The maze task
The data come from a series of collaborative, computermediated maze experiments. The setup is a modied
version of the original maze experiments conducted by
Garrod and Anderson (1987) and Garrod and Doherty
(1994), and is the same as that described in Healey and
Mills (2006). This task is ideal as it presents participants
with two kinds of coordination problem. First, it presents
participants with the recurrent need to refer to spatial locations, requiring participants to coordinate on the semantics of their referring expressions. Second, participants
are faced with the procedural coordination problem of
establishing routines for guiding each other through the
mazes.
The computer-mediated version of the maze task exhibits similar local and global patterns of coordination to
those observed in the original, spoken task (Healey & Mills,
2006; Mills 2007; Mills & Healey, 2006).
3.2. Procedure
2
The term iteration is too strong for describing most activities, as
their boundaries will be much less clear-cut, especially at low levels of
coordination. In the maze task it can be determined precisely when a
dyad starts and nishes each of the 12 mazes, hence the use of this
stronger term.
3
Familiar dyads do not perform better than unfamiliar dyads when
encountering novel referents in novel tasks (Schober & Carstensen, 2010).
163
Fig. 1. Example pair of maze congurations. The solid black circle shows the players current position, the cross represents the goal point that the player must
reach, solid bars are the gates, and the shaded areas are the switches which open the gates.
3.3. Data
The transcripts (see Appendix A, below) are selected
from a set of 32 dialogues as representative of the coordination problems encountered by participants. Each excerpt
is a full transcript of a single maze trial.
1. An initial stage of adapting to the new activity and

conversational partner.4
2. An intermediate stage of successful interaction, having
coordinated both on a referring schema as well as routines for managing the activity.
3. A nal stage of developing and systematizing the
communication system.
4. The development of semantic coordination

4.1. The global pattern to be explained
Over the course of the interaction, participants referring
schemas become progressively concise and systematized.
Consider the transcripts of Dyad 1, Trials 1, 5 and 11 (in
Appendix A, below). Each transcript contains the full dialogue
of solving an entire maze. Table 1 illustrates this pattern.
Note how the type of description changes over the course
of the interaction. Initially, descriptions use highly salient
features of the maze, e.g. sticking out cluster or large block
of squares. During the interaction, the descriptions, in
particular the counting conventions, become increasingly
systematized. Crucially, this development of coordination is
not reducible to referential contraction (Healey, 1997). The
descriptions used at the end of the experiment are not
abbreviated versions of those used initially: 5,7 is not an
abbreviated form of next to the sticking out cluster at the top.
A consistent nding is that initially participants take
over 100 turns to solve each maze. With task experience,
participants take fewer and fewer steps to solve each maze,
until the most coordinated pairs take about 8 turns to solve
a maze of equal difculty. What could account for these
patterns of coordination?
4.2. Analyzing the development of coordination in 3 distinct

stages
To analyze how coordination develops it is helpful to
decompose the development of semantic and procedural
coordination into 3 stages.
One caveat, this paper is not suggesting these are

distinct stages that can clearly be delineated, or indeed that
they are cognitively real. However they are useful heuristically to demonstrate interlocutors differential usage of
dialogue mechanisms, in particular repetition, that occur at
low vs. high levels of coordination.
4.3. Stages in the development of semantic coordination
4.3.1. Low levels of coordination
Initially, participants are faced with the problem of
adapting their linguistic resources to the task and to each
other. In tasks which prevent participants from using preexisting conventions, it is unsurprising that participants
encounter difculty in conventionalizing novel formmeaning pairings. It is perhaps more surprising that difculties are observed even when participants already possess
an ontology which should, on the face of it, be sufcient for
individuating locations in the mazes. Describing locations
appears deceptively simple. Most (all?) people know words
such as square, box, or row, and can also count from 1 to
7 (the maximum number of nodes in any maze). However,
participants must coordinate on the semantics of these basic
terms, e.g. whether rows can be diagonal, whether a
grouping of nodes should be called a square, whether lines
are the same as rows, and also which counting schema to use.
Consider Dyad 1, Trial 1, which demonstrates how
having words for individuating referents is no guarantee of
4
This stage is frequently ignored by studies of dialogue. Initial trials are
simply discarded as noisy practice trials.
164
successful interaction. Line 9 contains the unwieldy and

highly ambiguous description the third row up and second
to the end block. Depending on whether counting starts
from 0 or 1, and whether second to the end counts from
the left or right, this description could pick out many
possible locations (it is the grey switch labelled 1 in Fig. 1).
In Line 10 top could be either of the 2 topmost squares,
and consequently across is also ambiguous, as it doesnt
specify whether to count from the left or right. Further,
Participant1 uses different conventions in the horizontal
count 2 across than in the vertical count down one to
refer to the location where the position marker is.
This conict over which counting schema to use lasts till
Line 31. Note that when Participant1 proposes a Cartesian
system do it graph / grid fashion this differs from the
horizontal counting system second to the end block used
previously in Line 9. Subsequently in lines 48, 3rd block
across counts differently again, and is not in Cartesian
format. The horizontal Cartesian value should be 6,
counted from the edge of the window, not from the leftmost extension of the maze structure. This description also
proves unsuccessful, leading at line 53, to Participant1
attempting a less restrictive way of counting the number of
blocks in the row, which is then subsequently reused in
Line 73 as part of a composite description that triangulates
on the desired location by counting the rows, the number of
squares in a row and the locations position within that row.
Transcript 1 shows how during a single trial participants
attempt to negotiate different counting schemes. Coordination is achieved via partial repetition of the constituent
elements of each others descriptions, before settling on a
hybrid scheme that neither participant originally proposed,
and is less systematic than the graph/grid scheme proposed by Participant1,
This transcript, which is typical of initial maze game
dialogues, also demonstrates the fundamentally opportunistic nature of interaction. It is not the case that interlocutors explicitly ground each utterance and only
proceed once full understanding has been reached. Interlocutors start with descriptions that are not sufciently
well-dened to be used successfully, and then triangulate,
via partial repetition, on a less restrictive schema for
referring to maze locations. Although this slippage between interlocutors is often characterized as a trial and
error approach (Kecskes & Mey, 2008), this is not the same
as hit and miss: When interlocutors attempt to explicitly
negotiate the semantics, more often than not these attempts prove less successful than tacit negotiation (see also
Pickering & Garrod, 2004).
Note also how the interlocutors are also fundamentally
concerned with establishing what the other SHOULD know.
Turns such as like normal people (Dyad 1, Trial 1, Line 51)
and USING ENGLISH (Line 67) demonstrate that right from
the outset of the interaction, participants mark their
epistemic territories (Heritage, 2012), treating failure to
understand Cartesian systems as evidence of uncooperativity (Steensig & Drew, 2008).
4.3.2. Intermediate coordination: rening and systematizing
At intermediate stages of coordination, as in Dyad 3,
Trial 3 which shows a dyad completing a maze in 30 turns,
participants coordinate successfully, while continuing to

systematize and rene their referring expressions. Here
too, repetition plays a central role in shoring up the
counting conventions: in Line 11, Participant10 seeks clarication with second column?, yielding a response that
retains non-systematized elements that identify salient
features of the maze very bottom. Similarly, the clarication request in (Line 27) 1st column on the right?, is
answered with last column, which also resolves counting
via salience.
Further, this pattern of coordination suggests that the
semantic coordination achieved by the dyad can be greater
than that of the individual: Participant9 only uses descriptions that consist of vertical columns, whereas
Participant10 alternates between both vertical and horizontal row descriptions. Throughout, Participant9 only
uses descriptions that consist of vertical columns, yet when
complemented with the horizontal row descriptions of
Participant10, the descriptions constitute systematic Cartesian descriptions.
4.3.3. High coordination
In the later trials, pairs combine their descriptions into
highly contracted and systematized Cartesian coordinate
schemas. Consider Dyad 4, Trial 10. All that remains on the
conversational surface are the actual counts of the schema;
a description such as 3, 7 is eliding a rich description
which could be paraphrased as:
3 across, counting from the left edge of the main window,
starting at 0, treating all empty space as if they were made
of squares, ignoring paths between gates, and counting 7
down from the top of the maze in the same fashion.
Importantly, as shown in Table 1, these highly elliptical
abstract descriptions are NOT the same as those that are
used successfully in the less coordinated stages; the
ordering of the axes might differ, so might the counting
schema (i.e. counting from 0 vs. 1 or counting from the left
vs. right). This raises the bar for theories of conventionalization. Instead of conventions being established in a
winner-takes-all fashion (Steels, 2003), successful use of
a new, more systematic convention can require that both
participants incorporate each others perspectives, as in
Dyad 3, Trial 3.
Moreover, Garrod and Doherty (1994) argue that at high
levels of coordination, the semantic model can acquire
normative status. Although Garrod and Doherty propose
that normativity arises as a consequence of the interaction,
the negotiation of epistemic territory concerning Cartesian
descriptions in Dyad 1, Trial 1 shows dyads treating Cartesian conventions as normative from the outset, even
before they can be used successfully.
4.4. Accounting for semantic coordination
The central difculty of accounting for the development
of semantic coordination stems from the conuence of
referential contraction and the development of systematic
representations. This leads to some basic questions concerning the role of repetition. Some studies, e.g. Ferreira
and Bock (2006) propose two distinct forms of repetition
(short vs. long term priming) that are constrained by interindividual memory mechanisms. Focussing on the intraindividual, interactive patterns of repetition in maze task
dialogues points towards the interaction itself placing
different constraints on short vs. long term repetition. To
illustrate these constraints, this section distinguishes between local, turn-by-turn repetition, and global repetition
that occurs over the course of the interaction.
The kinds of repetition that occurs at both timescales
suggest that global convergence does not arise straightforwardly out of local repetition. Instead, local repetition is
best conceived as scaffolding which supports the repair of
existing representations as well as the construction of
novel representations. While some of these supporting
structures might become integrated into the resulting
representation, they need not be; some may be reused to
construct other representations, and still others may simply be used on a single occasion and then discarded.
4.4.1. Global repetition
At the end of the task, the most coordinated dyads
consistently use different semantic models than at the start.
Consequently, as interlocutors become coordinated they
repeat less and less of the initial semantic model. In addition, the global change in semantic model has a topdown
effect of accelerating lexical change. As identied by
Pickering and Garrod (2004), when participants switch to a
different semantic model, they also use different words to
identify the same constituent elements in the new model,
for example switching from talking about the top row to
talking about the 3rd line.
Instead of driving coordination, long-term repetition of
semantic model might simply be indicative of low coordination, caused by interlocutors being unable to move
beyond the local minimum in the coordination equilibrium
established at the start of the interaction. For example, if
participants are unable to establish counting conventions,
and instead continue using less systematized descriptions
that rely on salient features of the maze (e.g. at the top of
the biggest section of the maze), this will yield much higher
global repetition scores.
What, if anything is being repeated long-term over the
course of the task? A simplied characterization of the
global development of coordination (i.e. Table 1) is that
initially participants use referring expressions that rely on
salience, such as the sticking out bit or the large square.
These are subsequently used to clarify the counting conventions of more systematic descriptions, such as third
along the row from the sticking out bit. These, in turn, yield
more systematic descriptions such as third along on the
2nd row, which interlocutors combine into more systematized Cartesian descriptions such as third row, fth column, before nally stabilizing as highly elliptical Cartesian
descriptions such as (3,5). Each stage in this process relies
on prior coordination on a less systematic representation.
4.4.2. Local repetition
Since global coordination involves systematic development of different representations, this differentiation must
also be evident at a local turn-by-turn level. Models which
explain convergence arising out of local repetition are
165
intrinsically ill-suited for explaining these patterns. The

global development of coordination (illustrated in Table 1)
can only be achieved if interlocutors do not fully repeat
each other. Moreover, since there is a consistent pattern of
developing more systematic representations, the elements
that are not repeated cannot be explained away as noise or
random drift.
Although studies of priming typically assume that immediate repetition is, all things being equal, indicative of
successful communication, local structural repetition is one
of the central ways in which misunderstanding is signalled.
Perhaps the clearest example of this is embedded correction (Jefferson, 1987; Saxton, 1997) which shows how interlocutors exploit the local syntactic structure of each
others utterances in order to perform contrasts that
highlight the problematic element. Consider Dyad 3, Trial 3,
Lines 713. In Line 7, Participant9 instructs Participant10 to
move to the location described as .second colmn bottom. This proves unsuccessful, and leads, in Line 11 to
Participant10 asking clarication with second column?,
yielding Participant9s response second column from
right..very bottom in Lines 1213. Note how in Line 12,
Participant9 reuses the original term right from Line 9,
while subtly changing the underlying semantics: In Line 9,
the switch is referred to as being on the right, whereas in
Line 12, the switch is now located by how far it is from the
the right.
Note here that this change in semantic model which
elucidates the counting convention is achieved tacitly via
interlocutors local repetition of each others descriptions.
Further, this change is not achieved via immediate repetition of right. The part of the description that is repeated
(second column) is used as scaffolding to identify Participant9s turn in Line 7 as the source of problematic understanding (see also Healey, Purver, King, Ginzburg, &
Mills, 2003; Purver, Ginzburg, & Healey, 2002; Schlangen,
2004), which prompts Participant9 to go back and repair
the description.
Thus, high local repetition might be due to instances of
clarication requests. However, presence of clarication
requests need not indicate low levels of coordination.
Clarication is indicative of participants noticing, signalling
and attempting to resolve problematic understanding, and
is more effective than other, more explicit forms of negotiation (Pickering & Garrod, 2004), and more effective than
no clarication at all.
Arguably it is precisely at these junctures in the dialogue, where interlocutors perform local repetition to
perform clarication, that the local and global processes
intersect. This raises the question of how local structural
repetition facilitates the development of more systematic
semantic models. Extending the scaffolding analogy, when
descriptions that focus on salient features of the maze are
used interactively to resolve miscommunication, counting
local repetition does not distinguish between scaffolding
and the structures being constructed. Further, once the
scaffolding is removed, as in the highly elliptical utterances,
this shows that the structure is more robust (i.e.
coordinated).
One promising means of investigating the point of
contact between local and global repetition is to track the
166
specic lexical, syntactic and semantic representations that

are introduced, claried, retained and subsequently elided.
This would allow determining which kinds of local repetition (i.e. ne-grained addition, substitution and deletion
of the constituent elements of interlocutors referring expressions) promote or impede coordination, and would
also allow determining which structures are retained
globally as a consequence of specic forms of local
repetition.
solve the maze without being fully coordinated, answering

in Line 103 to what did you do differently? with nothing.
Similarly in Dyad 5, Trial 1. Lines 27. Participant6 requests do what you did before. Participant6 doesnt even
know which move to request. More importantly, neither
does Participant5, who asks in Line 8: now?. Intriguingly,
this shows that both participants are orienting towards a
nascent complementary structure, consisting of two parts,
neither of which is known to either participant.
5. The development of procedural coordination
5.2. Medium levels of coordination: constructing an activity
It should already be apparent that in addition to semantic coordination, the procedural language used to coordinate the timing and sequencing of participants turns
and actions also becomes progressively contracted.
Returning to Table 1, initially interlocutors use lengthy,
elaborate instructions, e.g. at (10 min): Tell me yours and
when I say, go onto..., which contracts after 40 min to
.yours?. Similarly, the instruction at (15 min) ...to open it
wait 3 seconds shortens to wait. .
It is helpful analyzing the interaction in 3 stages. From
the outset, participants are oriented towards establishing
complementary structures at specic junctures within
the activity. As coordination develops and the activity becomes sufciently well-dened, the complementary
structures become conventionalized as normative procedural conventions.
After achieving basic coupling of their actions and turns,

participants become progressively coordinated, developing
a stronger sense of the activity, i.e. the synergy (Fusaroli
et al., 2014) between them. Participants negotiate the
sequential implicativeness (Schegloff & Sacks, 1973) of their
contributions; that is, they negotiate what they do and
dont need to conrm, inform and request of each other,
and also establish the relevant junctures (Schegloff, 2007:
p223) in the unfolding activity where particular contributions need to be made.
There appear to be a whole range of communicative
devices used by participants to cement routines at specic
temporal and sequential junctures in the activity. These
devices work by reference to a particular sequential location within the activity as a whole. Put differently, the
dyads treat the procedural coordination problems they
encounter, not as one-off coordination problems, but as a
recurrent problem which occurs on each occasion of the
activity.
Consider Dyad 4, Trial 8, Line 6, which is from a medium
coordinated pair. Here ParticipantB tells participantA If I do
nothing it means I cannot get there. Importantly, ParticipantB does not simply say I cannot get there. This transcript is especially interesting: First, it demonstrates the
explicit introduction of a complementary convention in
which silence and inaction become conventionalized as a
move in the activity, within an adjacency pair whose rst
pair part is the ABSENCE of action. Second, conventionalization is achieved by giving a PRE-EMPTIVE form of accountable
absence (Schegloff, 2007) suggesting that there is a
normative dimension to this convention. Third, this turn is
introduced by reference to the next iteration of the activity.
Finally, this example underscores the fundamental importance of looking at multiple iterations of the same activity,
as one could only know whether silence had an interactive
function by looking at multiple prior occasions of the activity, where the convention was introduced. This also
suggests that the conversation analytic next turn proof
procedure (Heritage, 1984) would need to be extended to a
prior occasion of the activity proof procedure.
Other examples of establishing junctures by reference to
the activity as a whole include constructions such as now
you need to do X (e.g. Dyad 3, Trial 3, Line 19) or Next time
do X, instead of simply saying do X. This presupposes that
participants are already sufciently coordinated in the activity to know what the next time will be. Other forms of
anchoring an utterance in a particular sequential location
in the activity include saying at this stage we do X or here
we do X instead of simply making the request. These
5.1. Initially: weak coupling of pairs of actions

Initially, participants are faced with the basic coordination problem of coupling (Fusaroli, Raczaszec-Leonardi,
& Tyln, 2014; Riley, Richardson, Shockley, & Ramenzoni,
2011) their individual actions and utterances in order to
perform the joint actions of guiding each other through the
mazes. One basic strategy is for participants to overtly assume complementary roles, e.g. in Dyad 1, Trial 0, Line 2:
me navigating and you driving or Line 41: Ill direct.
At a ner grain, interlocutors turns are often oriented,
right from the outset, towards setting up local complementary structures produced by both participants: u go
rst, where do u need me to go? (Dyad 7, Trial 1, Line 3) or
because then I let you rst then Ill follow (Dyad 6, Trial 1,
Line 8).5
A basic format for accomplishing this is You do X and I
do Y or Ill do X and then you do Y, or Do X so I can do Y.
These differ from straightforward requests, as the requests
include both the rst and second pair parts of a nascent
adjacency pair, as opposed to asking the other simply to do
X.
Note that this process of coupling contributions is highly
opportunistic. Interlocutors do not have a clear notion of
what they are doing, much less a global plan of how to
proceed. See, for example, Dyad 1, Trial 1, who manage to
5
This example is of particular interest, as both participants are coordinated semantically, using Cartesian descriptions, but still encounter
difculty coordinating on the sequential structure of their interaction.
devices tend to be deployed towards the middle of the task,

once a certain level of coordination has been achieved.
Importantly, once these coordination problems have
been articulated, addressed, and resolved, they no longer
need to be mentioned explicitly. Since the solutions are
cemented in place, they too can be elided, allowing more
tacit and contracted procedural expressions, yielding an
activity with greater progressivity.
5.3. High levels of coordination

After establishing successful coordination, coordination
continues to develop. Participants turns become more
contracted, and also take fewer exchanges to solve the
mazes. Although each maze can be solved in a few steps,
each dyad establishes (often subtly) different complementary structures for signalling in the task. See for
example Dyad 4, Trial 10. Here, the participants signal to
each other how to progress through the maze by using
three words open, trapped and home. Importantly,
open does not simply mean that the gate is open. In much
the same way that the participants develop highly abbreviated referring conventions, open means something akin
to:
My gate has opened but I cant get to my goal, if you are
able to, please go to your goal and then tell me when youre
on it, otherwise tell me and I will then go to a switch that
you just mentioned and wait there either until one of my
gates opens or until you tell me to go to the goal.
Further, since this is the same Dyad described in the
preceding section which gave a pre-emptive form of
accountable absence (Dyad 4, Trial 8, Line 6), it would also
have an additional meaning of ..but if I do nothing it means
I cannot get there.
Importantly all this is achieved with a single word,
open, without an overt request, that simply articulates a
state in the activity. Note that trapped and home have
similarly rich sequentially implicative functions. Even utterances as innocuous as me now in Dyad 4, Trial 12, Line
6 are highly dyad-specic: Does it mean now its my turn to
speak, its my turn to guide you through a gate, my turn to
go onto a switch or my turn to wait. In much the same
way that different dyads might mean different things by
row or square, the same words used by different dyads
to coordinate procedurally acquire different procedural
functions that depend on the specic conversation history
and coordination problems encountered during the
interaction.
In the most coordinated pairs, each contribution effectively becomes a move in an activity with a multiplicity of
retrospective and prospective functions, resulting in an
extremely rich procedural semantics for coordinating the
activity. Each contribution can perform multiple simultaneous speech acts, i.e. participants are simultaneously
posing and answering multiple Questions Under Discussion (Ginzburg, 2012), often with a single word (See
Gregoromichelaki,
Cann,
&
Kempson,
2013;
Gregoromichelaki, Kempson, & Cann, 2012). Note that
although many questions are being asked (and answered)
167
with each turn, the turns do not have the canonical syntactic structure of a question.
A further hallmark of this coordinated stage is that the
task actions themselves (i.e. the opening and closing of
gates) also acquire sequentially implicative functions. In
the initial stages of the maze game, opening gates is typically preceded with a turn similar to can you go to my
switch on the top row, 5th square?. Within speech act
theory (Austin, 1962; Searle, 1969), this would be conceived
of as a perlocutionary effect, or as the second pair part of a
"projective pair" (Clark & Krych, 2004). However, towards
the end of the task, participants use the opening and
closing of gates to signal to each other whether or not they
are able to progress unproblematically through the maze
(as suggested by If I do nothing it means I cannot get there
in Dyad 4, Trial 8), and can be used as the rst pair part of a
complementary structure. Opening of gates becomes an
illocutionary act, instead of a perlocutionary effect.
5.3.1. Normativity and procedural pacts
How do interlocutors achieve this level of coordination?
Comparing dialogues in the early trials with the late trials is
all the more striking, given that all mazes are randomly
generated and are therefore, on average, equally difcult. It
appears that once a basic level of coordination has been
achieved, each procedural coordination problem, once articulated and resolved, no longer needs to be overtly
mentioned, consequently enriching the sequential implicativeness of each turn. The procedural function becomes
transparently absent, having disappeared from the
conversational surface. The data strongly suggest that in
the most coordinated dyads, this culminates in normative
conventions. Analogously to Brennan and Clarks (1996)
conceptual pacts, interlocutors are also establishing procedural pacts with each other.
Dyad 4, Trial 8 already showed participants explicitly
establishing a conventional meaning for silence at a key
juncture in the dialogue. Further evidence for procedural
pacts can be seen at high levels of coordination. Consider
Dyad 8, Trial 6: Note, in Line 5, Participant16 explicitly introduces ATG to mean at [your] goal, followed by a
respecication of ATG as a question, effectively asking
are you at your goal? which is subsequently recast as
AYG. Four trials later (Dyad 8, Trial 10), the dyad has
developed a much richer system, using AMG to mean At
my goal and AYS in Line 12 to ask the question Are you at
your switch?, as well as GC in Line 14 to mean gates are
clear. Of central interest is that checks (AYS), and statements of being at a particular location (AMG) both
become integrated into the same system of moves in an
activity.
It is perhaps surprising that, in a task that can be solved
in a few steps, participants develop (subtly) different ways,
not simply of solving the high-level task structure, but also
of coordinating how the activity unfolds, e.g. AYS and
CG in Dyad 8, vs. open?, goal?, exit?, done of dyad
4. These have different sequential imports that depend on
the specic interactional histories of each dyad.
A simple analogue is the utterance check in a chess
game, which is as much a move as physically moving a
chess piece. Similarly in the maze game uttering AYS and
168
opening or closing the others gates both constitute moves

in the activity.
Note that the contraction seen in AYS is distinct from
the contraction observed in spatial descriptions. Moreover,
since these forms are introduced explicitly as a convention,
and then subsequently adopted and used successfully, this
suggests they are stronger, more normative conventions.
5.4. Accounting for procedural coordination
Accounting for procedural coordination faces similar
problems to semantic coordination: the more coordinated
the dialogues, the less structure there is on the conversational surface, and the conventions also change during the
interaction.
Further complicating matters is that procedural coordination is underpinned by complementary structures.
Consider Dyad 4, Maze 10 which has reached ceiling levels
of coordination. The dyad solves the maze in a sequence
consisting solely of three terms: open, trapped and
home. Each turn performs a complementary next relevant contribution (Clark, 1996). So, although participants
have converged on a very small set of routines, their successful, coordinated use involves NO immediate, local, turnby-turn repetition of surface structure. From this perspective, once high levels of coordination have been achieved,
any increase in local, turn-by-turn repetition of structure
could be indicative of a decrease in forward momentum, and
consequently be indicative of a decrease in coordination.
This raises the bar for accounting how coordination is
established, as any account must show how systematic
differentiation of turns develops as the surface structure
becomes progressively elided. The simple surface form
obscures many layers from previous iterations, and consequently the same word, e.g. open acquires a very different
meaning at the end of the task than at the start. Thus longterm repetition between the start and end of the interaction is different from short term repetition of structure at
low vs. high levels of coordination.
Finally, as shown in Dyad 9, Maze 10, highly coordinated
participants can unproblematically interleave activityspecic utterances with everyday conversation. This raises the intriguing possibility that highly coordinated taskoriented dialogue should exhibit patterns of repetition
that are comparable to those of everyday conversational
interaction.
6. Putting the two together
Wittgenstein famously exhorted that the meaning of a
word is given by its use within a language game, stating
that an intention is embedded in a situation
(Wittgenstein, 1958, PI 337). The transcripts show how
interlocutors rapidly embed referring intentions and intentions associated with coordinating how the activity
unfolds. In semantic and procedural coordination, dyads
use the same words in different ways due to having
developed different language games for solving the mazes.
Even within dyads, the same words acquire meanings that
differ radically from their meanings at the start of the
interaction.
The data show that procedural function and reference

become intrinsically linked. In much the same way that it is
possible to ask someone to open a door by saying either
open? or door?, highly coordinated dyads develop
conventions that either use terms associated with the referents or the routines, e.g. using G5 in Dyad 9, Maze 10,
Line 20 or trapped in Dyad 4, Trial 10, Line 4 to instruct the
other participant to traverse a switch. These different coordination solutions that crystallize either referentially or
procedurally can be arrived at via very different trajectories, yet full an equivalent communicative function. It is
an open question which communicative pressures result in
either form of convention.
Further, within each dyad, participants often use
different terms to refer to the same maze locations if the
reference is embedded within a turn that has a different
conventionalized procedural function. In the original maze
game studies, Pickering and Garrod (2004) found that
when switching to a different semantic model with
different counting conventions, participants would use
different names for the same referent, example, e.g.
switching from the second row down from the top to the
third longest line. A similar phenomenon occurs with
procedural terms, see for example in Dyad 7, Trial 10. Here,
participants use home and exit? to refer to exactly the
same location (Both refer to the crosses in Fig. 1 which are
the goal locations that participants need to be on to exit the
maze). Here, home means something akin to Im on my
goal, can you get to yours?. This is then followed, in Line 8,
with a question exit?, which is a request for the location
of the goal, i.e. where is your exit?. The form of the
referring expression for exactly the same referent depends
on the particular routine in which it is embedded.
6.1. Interleaving unconstrained dialogue with a joint activity
Let us return to the myriad ways in which dialogue is
seamlessly interwoven with everyday activities, e.g.
conversing while buying items in a shop or while
walking down the street with a friend. These examples
of everyday, naturalistic, unconstrained interaction might
appear to bear little resemblance to the conned and
highly structured maze task (or indeed task oriented
dialogue in general). However, unconstrained conversational interaction in everyday activities depends intrinsically on interlocutors already being sufciently skilled
at coordinating how the activity unfolds, allowing the
conversation to be interleaved with activity-specic actions and utterances. By contrast, the primary focus of
dialogue between unskilled participants would necessarily be on establishing coordination, curtailing openended conversation.
In this light, consider Dyad 9, Trial 10, which shows two
participants engaged in unconstrained dialogue about
where to eat, and whether to irt with one of the staff.
Without knowing how difcult it is to establish coordination in the maze task, this excerpt seems unremarkable.
Notice here how the activity of solving mazes has receded
into the background, how few contributions are concerned
with the maze task, and how these contributions are
seamlessly interwoven with an unconstrained, everyday,
playful conversation. The transitions between conversing

and solving the maze are unmarked. There are no overt
signals for switching between the conversation and the
activity of solving the maze. Compare the ease with which
they are solving the maze with Dyad 1, Trial 1. There is
barely any mention of what each turn is doing; the procedural function of each turn has disappeared from the
conversational surface.
Arguably this transcript is more representative of a
conversational interaction where it just so happens that
they are playing a maze game with each other. At this
highest level of coordination, where interlocutors are able
to seamlessly switch between conversation, activity and
back again (and even play with the boundary between the
two), the dialogue bears a strong resemblance to everyday
conversation, say at a dinner table, between friends who
joke and gossip while simultaneously passing plates
around the table and proffering food to each other (Imagine
replacing the spatial such as a3 or d5 with the names of
various dishes, and questions such as are you on X yet?
with do you want more?).
Consider the question g5? in Line 20, or the question
salt?, asked at a dinner table. The answers to both questions are second pair parts of activity-specic adjacency
pairs. Being able to successfully perform these adjacency
pairs presupposes a background of having practiced multiple prior occasions of a joint activity, in which multiple
coordination problems were encountered, resolved and
conventionalized. Such prior coordination is intrinsic to
being able to converse open-endedly in unconstrained
everyday dialogue in joint activities.
7. Conclusion
The original maze task of Garrod, Anderson, and Sanford
(1984, 1987) was designed to unproblematically elicit
referring expressions from participants. However, it transpired that the referential coordination phenomena were
much more complex than originally anticipated. Similarly,
even though the task presents participants with a simple
procedural coordination problem (as evidenced by the
brevity of dialogues in the later stages), the interactive
development of idiosyncratic elliptical procedural conventions is equally complex.
One of the basic issues this raises is methodological. In
referential studies, patterns of interaction are typically
taken as indexing the communicative difculties associated
with securing reference. The ndings discussed here suggest that some of these patterns are primarily due to participants developing procedural coordination with each
other during the task.
Further, the phenomena discussed here emphasize
the importance of investigating the development of coordination between participants. Instead of treating
conversational interaction as a at static slice, what
needs to be explained is how coordination is established
and continues to be rened. Frequently studies engage
participants in a practice stage which is subsequently
excluded from analysis. The ndings discussed here
suggest that any model of dialogue must account for
how coordination develops during this initial practice
169
stage. Although more formal approaches to coordination

often presuppose that moves in a joint activity such as
request(x) or instruct(y) are somehow already given,
there is no reason to suspect that participants have a full
idea of what they or their partner are doing; initially in
the practice stage, participants often require many
interactive exchanges before establishing which act has
in fact been performed.
To conclude, the picture of dialogue that emerges is that
interlocutors do not have recourse to an Archimedean
vantage point outside the vagaries of language and the joint
activity from which error-free, successful communication
can be guaranteed. However, interlocutors constantly adapt
to each other via highly interactive forms of externalized
inferencing, in order to resolve the semantic and procedural coordination problems associated with the activity.
These solutions rapidly become conventionalized and
continue to be rened tacitly during the interaction into
idiosyncratic, highly efcient, possibly normative, vocabularies and routines, laying the foundation for unconstrained, naturalistic interaction.
Acknowledgements
This work is supported by Marie Curie IOF Fellowship:
PIOF-GA-2009-236632-ERIS. I am grateful to Herb Clark,
Pat Healey, Eleni Gregoromichelaki, Arash Eshghi and the
anonymous reviewers of this paper for valuable comments
and suggestions.
Appendix A
Dyad 1. Trial 1.
Participant1
Participant1
Participant1
Participant2
Participant1
Participant2
Participant1
Participant2
Participant1
1
2
3
4
5
6
7
8
9
Participant1
Participant2
Participant2
Participant2
Participant1
Participant1
Participant2
10
11
12
13
14
15
16
Participant1
Participant2
Participant1
Participant2
Participant1
Participant2
Participant1
Participant2
Participant1
Participant2
Participant2
Participant1
Participant2
17
18
19
20
21
22
23
24
25
26
27
28
29
Babs
me navigating and you driving
oh no
oh no@
When do we start
NOW
lemme direct you to my switch?
okey pokey
look at the third row up and second
to the end block
i am, from the top, 2 across and down one
so the middle block?
of the third row
???
nah
start counting from the bottom
think rows and block numbers from left
to right
6 along three up
rows from the top!
from bottm
why make it more complecated!
just do it
row:? block?
ok atrat again
i dnt no what ur saying dunb bum
start again
ok
row and block?
if you look at the screen
im looking...
(continued on next page)
170
(continued)
(continued)
Participant1
Participant1
Participant1
Participant2
Participant2
Participant1
Participant1
Participant2
Participant2
Participant1
Participant2
Participant2
Participant1
Participant2
Participant1
Participant1
Participant2
Participant2
Participant1
Participant2
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Participant1
Participant2
Participant2
Participant1
Participant2
Participant1
Participant1
Participant2
Participant1
Participant2
Participant2
Participant2
Participant2
Participant2
Participant1
Participant2
Participant2
Participant2
Participant1
Participant2
Participant1
Participant2
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
Participant1
Participant1
Participant1
Participant1
Participant2
Participant1
Participant2
Participant1
Participant2
Participant2
Participant1
Participant2
Participant2
Participant1
Participant1
Participant2
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
Participant2
Participant1
Participant1
Participant2
Participant1
Participant1
Participant2
Participant2
Participant1
Participant2
Participant1
88
89
90
91
92
93
94
95
96
97
98
then look at the bottom left block

then count in a normal graph/grid fashion
regardless of the maze
OMG!
why so comlicated!
count 6 along reading frm left to right
then three up
ROW!
THEN BLOCK NUMBER!
cant you understand my directions?
NO
ill direct
ok its row three from the bottm
my rst switch...
ok
ok
row 4
3rd block across
from bottom?
NO DUMB BUM i already told you,
from the TOP
ok
like normal people
are you there?
row 4 has three blocks
yes! well done
the third one across?
i have a gate before it
have you done it?
knob
ok
the other switch is...
7th row,
3rd block across
YAY!
im there
im at my goal!
wheres ur switch now,
USING ENGLISH
now you got to go to my switch u selsh cow
shut up bitch
from the top
if we were doing it your way, we wouldnt
even be here!
5 rows down
there are 6 on that row
its the 5th
reading from L to R
is it open?
no
wheres the other switch?
only one switch
just type row and block
just two numbers cant be that hard
from the top its the 5th row yeah
row 5
then it the 5th out of six
on that row
I HAVE gone over that switch
ALREADY!!!!!!!!!!!!!!!!!!!!!!
where is your other damn switch!!!!!
only one switch u prat
listen and stop trying to b awkward
row 5, block 5?
its block 5 out of 6 yeah
?
im doing that!
its not working
it went away and tehn came back
there now?
you walked past it
Participant2
Participant1
Participant2
Participant1
Participant2
99
100
101
12
103
there?
yay
I WAS RIGHT THERE
what did you do diff
nothing
Dyad 1. Trial 5.
Participant2
Participant1
Participant1
Participant2
Participant2
Participant1
Participant1
Participant2
Participant2
Participant1
Participant1
Participant2
Participant2
1
2
3
4
5
6
7
8
9
10
11
12
13
switches...
1,1 and 7,2
that it
this ones more complicated
where are ur gates
there at the barrier of 4,1
and 5,1
is that it?
and ur goal
and 6,4 and 7,2
my goal is 2,1
can you get to 3,1 and stay there please
and now to ur goal
Dyad 1. Trial 11.

Participant1
Participant2
Participant1
Participant2
Participant1
Participant2
Participant2
Participant1
Participant2
1
2
3
4
5
6
7
8
9
2,2 4,1 6,2

go to 2,1 please
ther
and now to ur goal
ther
shit
back to 2,1
ther
now to the goal
Dyad 3. Trial 3.
Participant9
Participant9
Participant10
Participant10
Participant9
Participant10
Participant9
Participant10
Participant9
Participant9
Participant10
Participant9
Participant9
Participant10
Participant9
Participant9
Participant9
Participant9
Participant10
Participant10
Participant9
Participant10
Participant10
Participant9
Participant10
Participant10
Participant9
Participant10
Participant9
Participant10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
ok wait
say all your switches
i have a one switch
it is the third row from the top
cant reach it
ok where is urs
mine : 1 - second colmn - bottom
now ?
on the right
no
second column ?
second colomn from right
very bottom
cant reach it
ok. the very top
on the lesft col
good
at the goal
now u need to go to the third row
u had it
what colmn?
last column from the right
third row
now?
no
thir row from the top
1st colm on the right?!
last column
now
yes
Dyad 4. Trial 8.
Participant1
Participant1
Participant2
Participant2
Participant1
Participant2
171
(continued)
1
2
3
4
5
6
5,4
6,6
3,1; 3,6; 5,2
Give me all urs
Thats it
If I do nothing it means I cannot get there
Dyad 4. Trial 10.

Participant1
Participant2
Participant2
Participant1
Participant2
Participant2
1
2
3
4
5
6
3,7 . 5,4
5,1 .. 7,6
open
trapped
open
home
Dyad 4. Trial 12.

ParticipantA
ParticipantB
ParticipantB
ParticipantB
ParticipantA
ParticipantA
1
2
3
4
6
7
3,1..5,6
4,2 5,4
open
home
Me now
home
Participant40
Participant50
Participant40
Participant50
Participant50
Participant40
Participant40
Participant50
Participant40
Participant50
Participant50
Participant40
Participant40
Participant50
Participant50
Participant40
Participant40
Participant40
Participant50
Participant40
Participant50
Participant40
Participant50
Participant40
Participant40
Participant40
Participant50
Participant50
Participant40
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
*one
open?
then i want you to go to 7
7 what?
this is gonna take a while
lol dude listen
right take note
ok
top left is 1,1
ok
so like chess
bottom right is 7,7
yes
ok
so where do u need me?
i want you to go to 7,3
7 is x axis
3 is y-axis
open??
got it?
yup
i am there
ok next?
i am in my goal
woohoo
where do i go?
ok i need u to go to..1,7
goal!
open sesame?
Dyad 5. Trial 1.
Dyad 7. Trial 10.
Participant5
Participant6
Participant5
Participant6
Participant5
Participant6
Participant6
Participant5
Participant6
Participant5
1
2
3
4
5
6
7
8
9
10
so go back to the switch

do weat u did b4 man
gey to the switch
do wat u did b4
is it open
no still closed
do wat u did b4
now?
no
tell me when
Dyad 6. Trial 1.
Participant26
Participant25
Participant25
Participant26
Participant25
Participant25
Participant26
Participant26
Participant25
Participant25
Participant25
Participant25
Participant26
1
2
3
4
5
6
7
8
9
10
11
12
13
oi
my gate is closed now, is urs open
i dont get it
is ur gate b4 that?
urs open now
?
look in the sequence is you gate b4 e6?
coz den i let u fru 1st den ill follow
my gate is my gate is b/w c 3 and c4
its closed now
is urs open
is urs open damit
yeh
1
2
3
4
5
6
7
8
9
dude
Hello there!
u go rst, where do u need me to go?
I have a gate to the last column
theres only one box there
open?
no
left, right or down?
if the left column is onr
Dyad 7. Trial 1.
Participant50
Participant40
Participant50
Participant40
Participant40
Participant50
Participant40
Participant50
Participant40
Participant40
Participant50
Participant40
Participant40
Participant40
Participant40
Participant50
Participant40
Participant40
Participant50
Participant40
Participant50
Participant50
1
2
3
4
5
6
7
8
9
10
11
12
13
7,7
blocked
3,3
you are so slow
yawn
home
shut up, im doing my best
exit?
/...
6,3
switch?
i need 3,7 7,7 or 6,4
go back
Dyad 8. Trial 6
Participant16
Participant17
Participant17
Participant17
Participant16
Participant16
Participant17
Participant16
Participant17
Participant17
Participant17
Participant16
1
2
3
4
5
6
7
8
9
10
11
12
1.2 and 3.7 and 7.6

5,1 and 1,4
can u reach mine
??
ATG-at your goal
ATG?
huh
AYG-at your goal
no im not i need u to open my gate
lol
ok u ATG
lol
Dyad 8. Trial 10
Participant16
Participant17
Participant17
Participant17
Participant16
1
2
3
4
5
lol how many more?

4,1 3,5
i dunno lol
im havin fun
2.5 and 3.2 and 6.7
(continued on next page)
172
(continued)
Participant17
Participant17
Participant16
Participant16
Participant16
Participant17
Participant16
Participant17
Participant17
Participant17
Participant16
Participant17
6
7
8
9
10
11
12
13
14
15
17
18
:)
there ya go
wheresd yours?
through my gate
AMG lol
4,1 and 3,5
AYS
nope u sure
GC
GO
AYS
AMG
Dyad 9. Trial 10.

Participant92
Participant80
Participant92
Participant80
Participant92
Participant80
Participant92
Participant80
Participant80
Participant92
Participant80
Participant92
Participant92
Participant92
Participant92
Participant92
Participant92
Participant80
Participant80
Participant80
Participant92
Participant92
Participant92
Participant92
Participant80
Participant92
Participant92
Participant92
Participant80
Participant92
Participant80
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
we did it again
ooops we did it again
that means less nandos from todays earnings
a3, d5
only 14 quid maan
damn i want more nandos !!!!!!
we shud come back in diguise
remember that chick from nandos
that really cute one,
b7 and g5
a3 or d5 mate
haha
yea
well
go fer her mate
she needs u
:D
I cant get to either
the X .. or the chick :P
g5 ?
chick maan
X aint gonna
get u nething
dude
are u on X yet ?
if we make it one hour and one min
we get paid for the extra one min effort lol
so go for it
were already been here for over an hour
hahaha
are you on X ?
References
Anderson, A. H., Bader, M., Bard, E. G., Boyle, E., Doherty, G., Garrod, S., et al.
(1991). The HCRC map task corpus. Language and Speech, 34(4), 351366.
Austin, J. L. (1962). How to do things with words. Oxford: Oxford University
Press.
Bakhtin, M. M. (1986). Speech genres and other late essays. In M. Holquist
(Eds.) (C. Emerson and M. Holquist, Trans.). Austin, Texas: University
of Texas Press.
Bangerter, A., & Clark, H. H. (2003). Navigating joint projects with dialogue. Cognitive Science, 27, 195225.
Brennan, S. E., & Clark, H. H. (1996). Conceptual pacts and lexical choice in
conversation. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 22, 14821493.
Brennan, S. E., Galati, A., & Kuhlen, A. (2010). Two Minds, one dialog:
coordinating speaking and understanding. The Psychology of Learning
and Motivation, 53, 301344.
Brown-Schmidt, S., & Tanenhaus, M. K. (2008). Real-time investigation of
referential domains in unscripted conversation: a targeted language
game approach. Cognitive Science, 32, 643684.
Clark, H. H., & Schaefer, E. (1989). Contributing to discourse. Cognitive
Science, 13, 259294.
Clark, H. H. (1996). Using language. Cambridge: Cambridge University

Press.
Clark, H. H., & Krych, M. A. (2004). Speaking while monitoring addressees for understanding. Journal of Memory and Language, 50(1),
6281.
Conchiglia, G., Rocca, G. D., & Grossi, D. (2007). On a peculiar environmental dependency syndrome in a case with frontal-temporal damage: Zelig-like syndrome. Neurocase, iFirst, 15.
Eneld, N. J., & Sidnell, J. (2013). Language presupposes an enchronic
infrastructure for social interaction. In D. Dor, C. Knight, & J. Lewis
(Eds.), The social origins of language: Studies in the evolution of language). Oxford: Oxford University Press.
Ferreira, V., & Bock, K. (2006). The functions of structural priming. Language and Cognitive Processes, 21(7-8), 10111029.
Fusaroli, R., Raczaszec-Leonardi, J., & Tyln, K. (2014). Dialogue as interpersonal synergy. New Ideas in Psychology, 32, 147157.
Fusaroli, R., Bahrami, B., Olsen, K., Frith, C. D., Roepstorff, A., & Tyln, K.
(2012). Coming to terms: an experimental quantication of the
coordinative benets of linguistic interaction. Psychological science,
23(8), 931939.
Galantucci, B. (2005). An experimental study of the emergence of human
communication systems. Cognitive Science: A Multidisciplinary Journal,
29(5), 737767.
Gann, T. M., & Barr, D. J. Speaking from experience: audience design as
expert performance. Language and Cognitive Processes, in press
Garrod, S. (1999). The challenge of dialogue for theories of language
processing. In S. Garrod, & M. Pickering (Eds.), Language processing).
Hove: Psychology Press.
Garrod, S., & Anderson, A. (1987). Saying what you mean in dialogue: a
study in conceptual and semantic coordination. Cognition, 27(2),
181218.
Garrod, D., Anderson, A., & Sanford, A. (1984). Semantic negotiation and
the dynamics of conversational meaning. Technical Report 1, Glasgow
Psychology Technical Report.
Garrod, S., & Doherty, G. (1994). Conversation, coordination and
convention: an empirical investigation of how groups establish linguistic conventions. Cognition, 53(3), 181215.
Garrod, S., Fay, N., Lee, J., Oberlander, J., & MacLeod, T. (2007). Foundations of representation: where might graphical symbol systems
come from? Cognitive Science: A Multidisciplinary Journal, 31(6),
961987.
Ginzburg, J. (2012). The interactive stanc. Oxford: Oxford University Press.
Gregoromichelaki, E., Cann, R., & Kempson, R. (2013). Coordination in
dialogue: subsentential speech and its implications. In L. Goldstein
(Ed.). Brevity Oxford University Press.
Gregoromichelaki, E., Kempson, R., & Cann, R. (2012). Language as tools
for interaction: grammar and the dynamics of ellipsis resolution. The
Linguistic Review, 29(4), 563584.
Healey, P. G. T. (1997). Expertise or expert-ese? The emergence of taskoriented sub-languages. In Proceedings of the 19th annual conference
of the cognitive science society. Stanford, California.
Healey, P., Purver, M., King, J., Ginzburg, J., & Mills, G. (2003). Experimenting with clarication in dialogue. In Proceedings of the 25th
annual meeting of the cognitive science society (pp. 539544).
Healey, P. G. T. (2004). Dialogue in the degenerate case? Peer commentary on Pickering and Garrod: the interactive alignment model.
Behavioral and Brain Sciences, 27(2), 201.
Healey, P. G. T., & Mills, G. J. (2006). Participation, precedence and coordination in dialogue. In Proceedings of the 28th annual conference
of the cognitive science society. Vancouver, Canada.
Healey, P. G. T., Swoboda, N., Umata, I., & King, J. (2007). Graphical language games: interactional constraints on representational form.
Cognitive Science: A Multidisciplinary Journal, 31(2), 285309.
Healey, P. G. T., Purver, M., & Howes, Christine (2010). Structural divergence in dialogue. In Proceedings of architectures and mechanisms for
language processing. York, UK, September 2010.
Heath, C., & Luff, P. (1992). Collaboration and control: crisis management and multimedia technology in London underground line
control rooms. Journal of Computer Supported Cooperative Work,
1(1), 2448.
Heritage, J. (1984). Garnkel and ethnomethodology. Cambridge: Polity
Press.
Heritage, J., & Raymond, G. (2005). Navigating epistemic landscapes:
acquiescence, agency and resistance in responses to polar questions. In J.-P. de Ruiter (Ed.), Questions). Cambridge University
Press.
Heritage, J. (2012). Epistemics in action: action formation and territories
of knowledge. Research on Language & Social Interaction, 45(1), 129.
Hutchins, E. (1995). Cognition in the wild. Cambridge, Mass.: MIT Press.

Jefferson, G. (1987). On exposed and embedded correction in social
interaction. In G. Button, & J. R. E. Lee (Eds.), Talk and social organisation (pp. 86100). Philadelphia: Multilingual Matters.
Kendon, A. (1976). The f-formation system: spatial-orientational relations
in face to face interaction. Man Environment Systems, 6, 291296.
Kecskes, I., & Mey, J. L. (2008). Intention, common ground and the
egocentric speaker-hearer. Mouton Series in Pragmatics 4.
Krauss, R. M., & Weinheimer, S. (1966). Concurrent feedback, conrmation and the encoding of referents in verbal communication. Journal
of Personality and Social Psychology, 4(3), 343346.
Larsson, S. (2007). Coordinating on ad-hoc semantic systems in dialogue.
In Proceedings of DECALOG.
Levinson, S. C. (1983). Pragmatics. Cambridge, UK: Cambridge University
Press.
Levinson, S. C. (1992). Activity types and language. In P. Drew, &
J. Heritage (Eds.), Talk at work: Interaction in institutional settings (pp.
66100). Cambridge University Press.
Linell, P. (1998). Approaching dialogue. Talk, interaction and contexts in
dialogical perspectives. Amsterdam: John Benjamins.
Louwerse, M. M., Dale, R., Bard, E. G., & Jeuniaux, P. (2012). Behavior
matching in multimodal communication is synchronized. Cognitive
Science, 36(8), 14041426.
Mills, G. J. (2007). Semantic coordination in dialogue: the role of direct
interaction. PhD Thesis. Queen Mary University of London.
Mills, G. J. (2011). The emergence of procedural conventions in dialogue.
In Proceedings of the 33rd annual conference of the cognitive science
society. Boston, USA.
Mills, G. J., & Healey, P. G. T. (2006). Clarifying spatial descriptions: local
and global effects on semantic coordination. In Proceedings of the 10th
workshop on the semantics and pragmatics of dialogue.
Mills, G. J., & Healey, P. G. (2008). Semantic negotiation in dialogue: the
mechanisms of alignment. In Proceedings of the 9th SIGdial workshop
on discourse and dialogue (pp. 4653). Association for Computational
Linguistics.
Mills, G. J. and Healey, P. G. T. The dialogue experimental toolkit, in preparation. The toolkit is available at http://dietsourcecode.sourceforge.
net.
Newman-Norlund, R. D., van Schie, H. T., van Zuijlen, A. M. J., &
Bekkering, H. (2007). The mirror neuron system is more active during
complementary compared with imitative action. Nature Neuroscience,
10, 817818.
Pickering, M. J., & Garrod, S. (2004). Towards a mechanistic psychology of
dialogue. Behavioral and Brain Sciences, 27(2), 169190.
Purver, M., Ginzburg, J., & Healey, P. (2002). On the means for clarication
in dialogue. In J. van Kuppevelt, & R. Smith (Eds.), Advances in
discourse and dialogue). KluwerAvailable at http://www.dcs.kcl.ac.uk/
pg/purver/publications.html.
Reitter, D., Moore, J. D., & Keller, F. (2006). Priming of syntactic rules in
task-oriented dialogue and spontaneous conversation. In Proceedings
173
of the 28th annual conference of the cognitive science society (pp. 685
690). Vancouver, Canada.
Riley, M. A., Richardson, M. J., Shockley, K., & Ramenzoni, V. C. (2011).
Interpersonal synergies. Frontiers in psychology, 2, 38.
Saxton, M. (1997). The contrast theory of negative input. Journal of Child
Language, 24, 139161.
Schank, R. C., & Abelson, R. P. (1977). Scripts, plans, goals, and understanding. Hillsdale, N.J.: Erlbaum.
Schegloff, E. A. (1992). Repair after next turn: the last structurally provided defense of intersubjectivity in conversation. American Journal of
Sociology, 97(5), 12951345.
Scribner, S. (1986). Thinking in action: some characteristics of practical
thought. In R. J. Sternberg, & R. K. Wagner (Eds.), Practical intelligence:
Nature and origins of competence in the everyday world (pp. 1330).
New York: Cambridge University Press.
Schegloff, E. A., & Sacks, H. (1973). Opening up closings. Semiotica, 8, 289327.
Schegloff, E. A. (1996). Conrming allusions: toward an empirical account
of action. American Journal of Sociology, 104, 61216.
Schegloff, E. A. (1998). Body torque. Social Research, 65(3), 535596.
Schegloff, E. A. (2007). Sequence organization in interaction: Volume 1: A
primer in conversation analysis. Cambridge University Press.
Schegloff, E. A. (2011). Word repeats as unit ends. Discourse Studies, 13(3),
367380.
Schlangen, D. (2004, April). Causes and strategies for requesting clarication in dialogue. In Proceedings of the 5th SIGdial workshop on
discourse and dialogue (pp. 136143).
Schober, M. F., & Carstensen, L. L. (2010). Does being together for years
help comprehension? In E. Morsella (Ed.), Expressing oneself/expressing ones self: Communication, cognition, language, and identity (pp.
107124) London: Taylor & Francis.
Schwartz, D. L. (1995). The emergence of abstract representations in dyad
problem solving. Journal of the Learning Sciences, 4(4), 321354.
Searle, J. R. (1969). Speech acts. Cambridge: Cambridge University Press.
Sebanz, N., Bekkering, H., & Knoblich, G. (2006). Joint action: bodies and
minds moving together. Trends in Cognitive Sciences, 10(2), 7076.
Simon, N. (1975). Echolalic speech in childhood autism. Consideration of
possible underlying loci of brain damage. Archives of General Psychiatry, 32(11), 14391446. PMID 812450.
Steels, L. (2003). The evolution of communication systems by adaptive
agents. In Adaptive agents and multi-agents systems (pp. 125140).
Springer Verlag.
Steensig, J., & Drew, P. (2008). Introduction: questioning and afliation/
disafliation in interaction. Discourse Studies, 10, 515.
Stivers, T., & Robinson, J. D. (2006). A preference for progressivity in
interaction. Language in Society, 35(03), 367392.
Sterponi, L., & Fasulo, A. (2010). How to go on: intersubjectivity and
progressivity in the communication of a child with autism. Ethos,
38(1), 116142. ISSN 0091-2131.
Wittgenstein, L. (1958). Philosophical investigations. Oxford: Blackwell.

New Ideas in Psychology: Gregory J. Mills

Uploaded by

Copyright:

Available Formats

New Ideas in Psychology: Gregory J. Mills

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

New Ideas in Psychology: Gregory J. Mills

Uploaded by

Copyright:

Available Formats

New Ideas in Psychology 32 (2014) 158173

Contents lists available at SciVerse ScienceDirect

New Ideas in Psychology

Dialogue in joint activity: Complementarity, convergence

Dialogue in joint activities is the most basic form of

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

Based on these insights, a vast body of work has uncovered

semantics of their referring expression to each other and

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

negotiation between participants (Clark & Schaefer, 1989;

task-oriented dialogue work against this straightforward

OK, my grey switchpoint which you have to get

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

momentum, interlocutors must necessarily do something

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

Dyads must solve 12 randomly generated mazes (see

2.3. How do semantic and procedural coordination develop?

Pairs of native English speaking university students

When interlocutors establish a novel communication

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

1. An initial stage of adapting to the new activity and

4. The development of semantic coordination

4.2. Analyzing the development of coordination in 3 distinct

One caveat, this paper is not suggesting these are

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

successful interaction. Line 9 contains the unwieldy and

participants coordinate successfully, while continuing to

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

intrinsically ill-suited for explaining these patterns. The

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

specic lexical, syntactic and semantic representations that

solve the maze without being fully coordinated, answering

5. The development of procedural coordination

5.2. Medium levels of coordination: constructing an activity

After achieving basic coupling of their actions and turns,

5.1. Initially: weak coupling of pairs of actions

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

devices tend to be deployed towards the middle of the task,

5.3. High levels of coordination

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

opening or closing the others gates both constitute moves

The data show that procedural function and reference

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

playful conversation. The transitions between conversing

stage. Although more formal approaches to coordination

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

then look at the bottom left block

Dyad 1. Trial 11.

2,2 4,1 6,2

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

Dyad 4. Trial 10.

Dyad 4. Trial 12.

so go back to the switch

1.2 and 3.7 and 7.6

lol how many more?

G.J. Mills / New Ideas in Psychology 32 (2014) 158173

Dyad 9. Trial 10.

Clark, H. H. (1996). Using language. Cambridge: Cambridge University

G.J. Mills / New Ideas in Psychology 32 (2014) 158173