姝 Academy of Management Review

2005, Vol. 30, No. 1, 58–77.

DISCOURSE AND COLLABORATION: THE ROLE

OF CONVERSATIONS AND COLLECTIVE
IDENTITY
CYNTHIA HARDY
University of Melbourne

THOMAS B. LAWRENCE
Simon Fraser University

DAVID GRANT
University of Sydney

We explore the relationship between discourse and interorganizational collaboration,

arguing that interorganizational collaboration can be understood as the product of
sets of conversations that draw on existing discourses. Specifically, we argue that
effective collaboration, which we define as cooperative, interorganizational action
that produces innovative, synergistic solutions and balances divergent stakeholder
concerns, emerges out of a two-stage process. In this process conversations produce
discursive resources that create a collective identity and translate it into effective
collaboration.

Organizations in all sectors of society increas- nizational relationship in which participants

ingly are becoming involved in a variety of col- rely on neither market nor hierarchical mecha-
laborative arrangements—alliances, partner- nisms of control to gain cooperation from each
ships, roundtables, networks, and consortia—in other (Heide, 1994; Lawrence et al., 2002; Milne,
order to promote innovation, enter new markets, Iyer, & Gooding-Williams, 1996; Phillips et al.,
and deal with intractable social problems 2000; Powell, 1990). This definition encompasses
(Dyer & Singh, 1998; Gray, 1989; Gulati, Nohria, a large variety of collaborative arrangements
& Akbar, 2000; Powell, Koput, & Smith-Doerr, and distinguishes them from relationships that
1996; Simonin, 1997). By collaborating, organi- are cooperative but where cooperation is pur-
zations hope to leverage the differences chased, as in buyer-supplier relationships, or
among them—in terms of knowledge, skills, where it is based on the invocation of legitimate
and resources—so as to develop innovative, authority, as in cooperation between a state reg-
synergistic solutions to complex problems ulatory agency and an organization operating
they cannot solve on their own (Bresser & Harl, within its jurisdiction.
1986; Gray, 1989; Gulati, 1999; Phillips, Lawrence, Although collaboration has the potential to
& Hardy, 2000). Thus, in successful interorgani- produce powerful results, not all collaborations
zational collaborations, individual participants realize this potential. Many collaborations fail
are able to draw on their organizational ties to produce innovative solutions or balance
and, at the same time, transcend those ties to act stakeholder concerns, and some even fail to
collectively (Astley, 1984; Lawrence, Hardy, & generate any collective action whatsoever
Phillips, 2002). (Gray, 1989; Huxham, 1996; Lawrence et al., 2002).
There is little agreement as to how to define Consequently, researchers and managers alike
collaboration; thus, a wide range of definitions are interested in identifying the factors and pro-
exists. We define it as a cooperative, interorga- cesses that lead to the accomplishment of effec-
tive collaboration, by which we mean collabo-
ration that (1) leverages the differences among
We thank Dev Jennings and three anonymous reviewers
for their insights and ideas on earlier versions of this article.
participants to produce innovative, synergistic
We also acknowledge the financial support of the University solutions and (2) balances divergent stake-
of Melbourne. holder concerns.
58
2005 Hardy, Lawrence, and Grant 59

Effective collaboration among organizations thus, provides a basis for practical insights into
is a difficult task; not only must cooperation and how conversations might be managed to in-
innovation be achieved, but the interests of crease the likelihood of effective collaboration.
those organizations represented in the collabo- In this way, we build on other scholarly exami-
ration must also be met. The challenge is even nations of the role of language in effecting or-
greater because cooperation among partici- ganized, collective action (Ford & Ford, 1995;
pants cannot be secured through market or hi- Heracleous & Barrett, 2001; Palmer & Dunford,
erarchical forms of control; although market or 1996).
hierarchical concerns may prompt the formation A second benefit of a discursive approach is
of a collaboration, and all collaborations occur that it highlights the processual and temporal
within the broad context of both markets and aspects of collaboration, thus allowing us to
hierarchies, these mechanisms do not operate
view collaboration as a social accomplishment
directly within the collaboration itself. Effective
that occurs in an iterative fashion over time. The
collaboration therefore depends on the relation-
focus on language in use means that collabora-
ships among participating members, which are
tion can be tracked more easily in a continuous
negotiated on an ongoing basis throughout the
life of the collaboration. Consequently, collabo- fashion by following conversations on an ongo-
ration represents a complex set of ongoing com- ing basis, rather than trying to ascertain mem-
municative processes among individuals who bers’ beliefs at discrete points in time.
act as members of both the collaboration and of Finally, a discursive approach facilitates the
the separate organizational hierarchies to development of theory and research that attend
which they are accountable. to the multiple levels on which collaboration
In order to address these complexities and occurs; collaboration involves individual partic-
examine how effective collaboration can be ipants working in collaborative teams while
achieved, we consider collaboration from a dis- representing the interests of organizational
cursive perspective (Lawrence, Phillips, & stakeholders. A discursive approach is attuned
Hardy, 1999). This allows us to develop a model to the complex interrelationships among these
of interorganizational collaboration that relates levels and how they impact collaboration be-
effectiveness to the conversations among partic- cause it examines the conversations that occur
ipants and to the wider discourses in which the within both the collaboration and the organiza-
collaboration takes place. In brief, our model tion, and because it explores how the language
suggests that effective collaboration results used in conversations draws on discourses op-
from a two-stage process in which participants erating at organizational and societal levels (cf.
engage in conversations that can include not Alvesson & Kärreman, 2002; Hardy & Phillips,
only face-to-face dialogue but also a variety of 1999).
other discursive practices, such as memos, let- We present our argument in four sections.
ters, e-mails, and minutes of meetings (Ford & First, we provide a theoretical overview of orga-
Ford, 1995). The first stage of our model high-
nizational discourse that highlights the role of
lights the importance of a discursively con-
conversations in generating discursive objects
structed collective identity in the achievement
that represent resources for action. Second, we
of effective collaboration. The second stage of
introduce the relationship between collective
the model shows how this collective identity is
translated through further conversations into in- identity and collaboration and discuss the forms
novative and synergistic action. Thus, by exam- of conversation necessary for the production of a
ining the discursive practices that constitute collective identity among organizational repre-
collaboration, we are able to explore the ways in sentatives in a collaboration. Third, we argue
which conversations can be managed to in- that collective identity alone is not sufficient to
crease the likelihood of effective collaboration. guarantee collaboration, and we explore the
Adopting a discursive approach to the issue of forms of conversation that provide the addi-
effective collaboration provides a number of sig- tional steps that translate identity into effective
nificant benefits. First, it directs attention to the collaboration. We conclude with a consideration
communicative practices among participants, of the implications of our model for research and
which are critical to effective collaboration, and, practice.
60 Academy of Management Review January

ORGANIZATIONAL DISCOURSE beyond any individual text from which it is com-

posed (Chalaby, 1996; Phillips, Lawrence, &
In this section we provide an overview of key
Hardy, 2004). Heracleous and Barrett describe
concepts from research on organizational dis-
the relationship between texts and discourse as
course that underpin our model of effective col-
analogous to that between action and social
laboration. A discursive approach to organiza-
structure: “Just as the structural properties of
tional phenomena is more than a focus on
social systems are, according to Giddens, in-
language and its usage in organizations. It
stantiated as social practices, so the structural
highlights the ways in which language con-
properties of discourse are instantiated in daily
structs organizational reality, rather than simply
communicative actions” (2001: 758). The implica-
reflects it (Grant, Keenoy, & Oswick, 1998; Phil- tion of this relationship is that researchers must
lips & Hardy, 2002). The study of discourse fo- examine sets of texts that describe and consti-
cuses on the ways in which actors draw on, tute organizational realities, as well as the com-
reproduce, and transform discourses, and, in so plex relationships among texts and among dis-
doing, produce a social reality consisting of dis- courses (Phillips & Hardy, 2002).
cursively constituted objects and ideas. Accord- One form that relationships among texts take,
ingly, research on organizational discourse ex- which is critical for this article, is that of a “con-
amines the processes through which discursive versation” (Taylor & Van Every, 1993). We define
objects are formed in organizations, the ways in a conversation as a set of texts that are pro-
which those objects constitute the social world, duced as part of an interaction between two or
and the consequences of those discursive pro- more people and that are linked together both
cesses and objects for organizations. temporally and rhetorically (Ford & Ford, 1995).
Three concepts are critical to our discursive Texts only exist as part of the same conversation
approach: discourse, text, and conversation. Al- if they are in some way responsive to each other,
though a wide variety of definitions have either directly or indirectly (a rhetorical connec-
emerged across different disciplines (van Dijk, tion), and are produced through chronologically
1997), we define a discourse as a set of interre- sequenced discursive acts (a temporal connec-
lated texts that, along with the related practices tion). Conversations are important in under-
of text production, dissemination, and reception, standing the role of language in organizations:
bring an object or idea into being (Fairclough, consequential action is not so much the result of
1992; Parker, 1992; Phillips & Hardy, 1997). For disconnected utterances or isolated texts but,
example, the discourse of business strategy has rather, is produced through ongoing linguistic
produced a series of ideas, such as “core com- exchanges among actors that draw on broader
petence” and “strategic planning” (Knights & discourses and produce discursive objects that
Morgan, 1991), that have shaped business prac- act as resources for action and for further con-
tice; the discourse of AIDS has established, in versations (Fairclough, 1992). Thus, conversa-
the developed world, the concept of a “patient- tions exist in a recursive relationship in which
activist,” which has led to various forms of ac- existing discourses provide resources to actors
tive political opposition to governments and who engage in conversations that in turn pro-
pharmaceutical companies (Maguire, Phillips, & duce, reproduce, and transform those discourses
Hardy, 2001). Discourses therefore help to consti- (Taylor et al., 1996).
tute a material reality by producing “identities, Ford and Ford (1995) argue that the forms of
contexts, objects of value, and correct proce- conversation associated with initiating organi-
dures” (Taylor, Cooren, Giroux, & Robichaud, zational change are those that establish a need
1996: 28), which lead to particular practices for change. From a discursive perspective, a
through the way they shape what can be said “need for change,” whether in the shape of an
and who can say it (Deetz, 1992; Fairclough, environmental shift, an organizational problem,
1992). or a political agenda, is a discursive object that,
Discourses are embodied in sets of texts that once produced, is available for use by other
come in a wide variety of forms, including writ- interested actors (Parker, 1992). In a collabora-
ten documents, speech acts, pictures, and sym- tive context, individual participants might draw
bols (Grant et al., 1998; Taylor & Van Every, 1993). on a need for change in conversations both in-
At the same time, discourse has an existence side and outside the collaboration in order to
2005 Hardy, Lawrence, and Grant 61

justify resource allocations, motivate fellow par- are discursively formed through conversations
ticipants and organizational colleagues, and le- in this context. In the following section we ex-
gitimate decisions of inclusion and exclusion in amine the conversations that affect the likeli-
the collaboration. As the need for change is hood that collective identity will lead to effec-
drawn on in other conversations, it may become tive collaboration (see Figure 1 for an overview
institutionalized in the discourse as one— or of the model). We discuss these two sets of dy-
perhaps the only—legitimate way of talking namics sequentially, but do not imply some sort
about the issue. of lock-step process; we separate them for ana-
These dynamics illustrate the potential politi- lytic purposes only, fully realizing they likely
cality of conversation: conversations are at least overlap in time and space.
partially the product of strategic action not only
by elites but by a plurality of individuals in-
Collective Identity and Action
volved in the production and consumption of
texts (Grant et al., 1998). Moreover, discursive The idea of collective identity is grounded in a
practices, including conversation, have the po- variety of traditional sociological concepts,
tential to produce significant political effects ranging from Durkheim’s “collective conscious”
that result in the differential distribution of ad- to Marx’s “class consciousness.” It “addresses
vantage among individuals and organizations the ‘we-ness’ of a group, stressing the similari-
(Fairclough, 1992; Mumby & Clair, 1997). ties or shared attributes around which group
In summary, we define a discourse as a set of members coalesce” (Cerulo, 1997: 386). In orga-
interrelated texts and their related practices of nizational research, collective identity predom-
consumption, production, and distribution, inantly has been explained as members’ conver-
which bring into being an object or idea. The gent beliefs about the central, enduring, and
texts that populate discourses range from writ- distinctive attributes of their organization (Al-
ten works to speech acts to nonlinguistic sym- bert & Whetten, 1985). These beliefs affect the
bols and images. Temporally and rhetorically way in which members interpret and react to
related texts constitute conversations in which issues facing the organization (Dutton & Duk-
participants draw on and simultaneously pro- erich, 1991; Gioia & Thomas, 1996) by influencing
duce discursive objects and ideas. Thus, a dis- the importance that members attach to them,
cursive approach to interorganizational collab- whether or not they see these issues as a threat
oration is inherently processual and contextual or opportunity, and whether they are willing to
(Lawrence et al., 1999): our focus on conversation invest time and energy in addressing them (Els-
highlights the processes through which collab- bach & Kramer, 1996; Gioia & Thomas, 1996).
oration is enacted over time; our interest in dis- Some writers argue that convergence around
course ensures a concern for the broader context collective identity leads to action by increasing
that provides the discursive objects that act as organizational commitment (Ashforth & Mael,
conversational resources for potential collabo- 1989; Dutton et al., 1994) and strengthening orga-
rators. nizational culture, motivating employees to
work cooperatively (Fiol, 1991; Saffold, 1988;
Schein, 1986). More recent work, however, high-
COLLECTIVE IDENTITY
lights the importance of tensions, fluidity, and
In the remainder of this article, we explore the paradox in organizational identity (Fiol, 2002;
relationship between collective identity and ef- Gioia, Schultz, & Corley, 2000; Pratt & Foreman,
fective collaboration and the role that discourse 2000).
and conversations play in that relationship. We adopt a discursive perspective, which
Central to our argument is the idea that discur- builds on recent interest in the tensions and
sive processes produce collective identities, fluidity associated with collective identity. From
which lead to various forms of collective action, a discursive perspective, a collective identity
potentially including effective collaboration. In exists as a discursive object produced in and
this section we discuss the fundamental role through conversations, rather than as a cogni-
that collective identity plays in achieving col- tively held belief (Taylor & Van Every, 1993).
lective action in the context of collaboration, When collaborating partners discursively pro-
and we then examine how collective identities duce a collective identity, they produce a discur-
62 Academy of Management Review January

FIGURE 1
A Model of Collective Identity, Conversations, and Effective Interorganizational Collaboration

sive object that refers to themselves as some Hogg, 2000; Tajfel, 1972), a discursive approach
form of collective, rather than as simply a set of does not examine what members are thinking,
disconnected individuals or as a group of orga- nor does it attempt to relate collective identity to
nizational representatives. This collective iden- individual social identities. Rather, its focus is
tity “names” the group—it gives it an identity primarily on the constructive effects of conver-
that is meaningful to its members and to its sations in which participants describe them-
stakeholders—and is shared, in the sense that selves as a collective. A discursive conceptual-
members collectively engage in the discursive ization of collective identity provides a powerful
practices that produce and reproduce it over basis for understanding the dynamics of collab-
time. Thus, our focus is on collective identity as oration, because it situates collective identity in
a linguistically produced object embodied in the language in use among members and
talk and other forms of text, rather than as a set avoids the need to assess the degree of conver-
of beliefs held in members’ minds. gence across the minds of individuals. It shifts
Discourse-oriented studies of collective iden- attention from the intentions and attitudes of
tity concentrate on the processes through which individuals to their observable linguistic prac-
a collective identity is produced via the creation tices and the effects of those practices on social
of texts, and on the relationship of collective relationships and action (Potter & Wetherell,
identity as a discursive object to patterns of ac- 1987).
tion. Unlike research on social identity (e.g., We argue that a collective identity, under-
Abrams & Hogg, 1999; Ashforth & Mael, 1989; stood as a discursive object, is critically impor-
2005 Hardy, Lawrence, and Grant 63

tant in achieving effective collaboration. The orations take on a mixture of the attributes of
discursive construction of a collective identity participating organizations. At a minimum, in-
enables participants to construct themselves, terorganizational collaboration represents a
the problem, and the solution as part of a col- complex arena in which to construct a collective
laborative framework in which the potential for identity, because it involves individual partici-
joint action is both significant and beneficial. A pants working together as a group while re-
collective identity also acts as a resource for maining connected to the organizations in
participants in future conversations—to ad- which they work. This question thus arises: How
dress, for example, internal conflict among is a collective identity produced in the case of
members or differences in organizational inter- collaboration?
ests that might otherwise jeopardize the collab- Collins argues that collective identities are
oration. It helps individual participants in the produced through conversations that discur-
collaboration attach importance to an issue, col- sively construct common “realities” regarding
lectively invest time and energy in it, commit to membership in different groups:
any compromises involved in tackling it, take For example, shop talk involves membership in
collective risks, and secure support from their occupational groups, political and other ideolog-
respective organizations. A collective identity is ical talk involves contending political coalitions,
particularly important in the case of collabora- entertainment talk invokes groups with various
tion, because the absence of market and hierar- tastes, general discussion invokes different intel-
lectual and nonintellectual strata, while gossip
chical controls among participants means that and personal talk invoke specific and sometimes
they cannot rely on monetary exchange or hier- quite intimate memberships. Again, it is not im-
archical authority to achieve cooperation. Thus, portant whether what is said is true or not, but
the discursive construction of a collective iden- whether it can be said and accepted as a common
tity through conversations can help participants reality for that moment—that is what makes it an
emblem of group membership (Collins, 1981: 999 –
in a collaboration identify with their partners 1000).
and provide a rationale for the cooperation that
is essential to its effectiveness (Gheradi, 1995). In the context of collaboration, this argument
This relationship between collective identity suggests that the types of conversations in
and effective collaboration leads to our first which participants engage have a critical effect
proposition. on the likelihood a collective identity will
emerge.
Proposition 1: Effective interorganiza- Two specific types of conversations are par-
tional collaboration will be more ticularly critical to the production of a collective
likely to ensue when the participants identity within a collaboration: (1) those conver-
in a collaboration discursively pro- sations that produce generalized membership
duce a collective identity for them- ties— discursively constructed relationships
selves. that connect participants to a common issue
around which the collaboration is organized—
and (2) those conversations that produce partic-
Conversations and Collective Identity
ularized membership ties, which connect the
Although a collective identity is critical to suc- participants directly to each other, rather than
cess, the complexities of interorganizational col- indirectly through an issue (Hardy, Lawrence, &
laboration make the discursive construction of a Phillips, 1998). Although we discuss these ties
collective identity relatively difficult to achieve: separately, conversations may involve an exten-
whereas some notion of the central, enduring, sive network of connected exchanges, and, con-
and distinctive attributes of an organization can sequently, both generalized and particularized
often be taken for granted (although perhaps in membership ties may be produced within the
a minimal or problematic way), collaborations same conversations. Together, generalized
are often neither central (participants’ home or- membership ties and particularized member-
ganizations continue to take priority) nor endur- ship ties provide the foundations for the discur-
ing (collaborations often have a fixed life span sive construction of a collective identity.
or lead a tenuous existence). It may also be Generalized membership ties. Generalized
difficult to develop distinctiveness when collab- membership ties revolve around membership in
64 Academy of Management Review January

a community based on a connection to some laypeople; however, it also constrained conver-

issue or problem (Collins, 1981). These ties are sations—for example, by restricting incompati-
produced through conversations that define an ble talk of “irresponsible” activism.
issue in a particular way and that connect orga- Generalized membership ties play a critical
nizations to that issue. Conversations that gen- role in the discursive production of a collective
erate generalized membership ties provide a identity in the context of collaboration by facil-
basis on which participants can identify an is- itating participants’ articulation of a shared is-
sue as relevant to their organizations and con- sue as important to a number of stakeholders,
sequently identify themselves as interested in including themselves, and as sufficiently endur-
or affected by it. ing to warrant some form of collective response.
The conversations that produce generalized For example, in a study of collaboration among
membership ties are those in which issues are employment service organizations, Hardy et al.
constructed, and so involve the production of (1998) found that only when conversations oc-
accounts of the world as problematic and requir- curred that explicitly built generalized member-
ing action (Lawrence et al., 1999). Such conver- ship ties around a common issue were partici-
sations might be diagnostic, in that they define pants able to articulate an enduring rationale
a problem and specify its causes; prognostic, in for a collaborative partnership and a collective
that they build consensus and mobilize action; vision of the future. Until then, a lack of gener-
or motivational, in the form of a general call to alized membership ties had impeded the pro-
arms to potential participants and stakeholders duction of a collective identity, despite monthly
(Benford & Snow, 2000). The way in which an meetings over an eighteen-month period: forced
issue is constructed has important political ef- together by their primary funder, and never hav-
fects on collaboration: it places limits on poten- ing discussed who they were as a group or what
tial interactions and plays an important role in their purpose was, participants had not clearly
determining who has a legitimate case for mem- articulated an issue that located them as mem-
bership in the collaboration (Gray, 1989; Gray & bers of a larger domain. The relationship be-
Hay, 1986). tween generalized membership ties and collec-
Some organizations will be more powerful tive identity leads to our next proposition.
than others in defining the issue (Altheide, 1988).
We do not suggest that these actors necessarily Proposition 2: Participants in an inter-
enter into these conversations with fixed or pre- organizational collaboration will be
determined views that they intend to “sell” (Dut- more likely to discursively construct a
ton & Ashford, 1993); rather, the discursive con- collective identity when they engage
struction of issues often occurs in an emergent in conversations that establish gener-
fashion as actors coproduce those constructions alized membership ties among them-
that affect the distribution of advantage and selves.
disadvantage, although in ways that were not
necessarily intended or predicted. Particularized membership ties. The produc-
At the same time, actors do not operate in a tion of a collective identity depends on partici-
vacuum but in a larger discursive setting that pants understanding themselves as tied to each
shapes these conversations; only conversations other directly, as well as indirectly through the
that are sensible and legitimate within the con- issue. Collins (1981) refers to ties that produce
text of prevailing discourses will produce gen- and reproduce direct relationships among con-
eralized membership ties. For example, in the versational participants as particularized mem-
development of a collaboration between Cana- bership ties. These ties describe status, author-
dian AIDS community groups and pharmaceuti- ity, and task-role relationships, as well as
cal firms to improve access to AIDS treatment affiliative and collegial relations. In a collabo-
(Maguire et al., 2001), community members drew ration, particularized membership ties involve
strategically on scientific discourse. This pre- routines, procedures, and structures that delin-
existing discourse provided the means by which eate, for instance, what roles individuals take in
to construct generalized membership ties and problem-solving or idea generation processes,
allowed community members to connect to the who attends particular meetings, and, more for-
issue of treatment as responsible and informed mally, who can legitimately make decisions or
2005 Hardy, Lawrence, and Grant 65

speak on behalf of the collaboration (Gray, 1989; While generalized ties provide a substantive
McCann, 1983). basis for the existence of a collective identity as
The conversations that produce particularized stakeholders collectively face a particular issue,
membership ties are those that refer to specific particularized ties provide a processual and
persons, places, and objects and, consequently, structural basis for specific participants to ad-
provide a set of discursive resources from which dress the issue in a particular forum. The col-
participants can position themselves as con- laboration in question is thus constructed as a
nected in specific, identifiable ways. Gricar and real and distinct entity, separate and different
Brown (1981), for example, describe collabora- from the organizations involved in it and more
tions as mapping relationships and interdepen- than simply a set of their representatives.
dencies among member organizations, as well In a study of a collaboration between an aid
as coordinating activities among these stake- organization located in the West Bank and a
holders. Through these conversations, particu- department of a Scandinavian university, Law-
larized membership ties are produced as certain rence et al. (2002) found that the two organiza-
stakeholders are included, patterns of interde- tions were able to discursively produce particu-
pendence are established, authority and status larized ties to one another, despite significant
are conferred, and roles and responsibilities are cultural and sectoral differences between them,
assigned. In addition, the repeated nature of as a result of extensive conversations among
routine conversational activity—about particu- their respective employees regarding the specif-
lar matters, between particular people, in par- ics of the project and their roles in it. These
ticular places— creates specific attachments included operational and information meetings
and relationships (Collins, 1981; Taylor et al., among employees of the two organizations, as
1996). well as a series of multiparty meetings that also
The discursive production of particularized involved government and nongovernmental or-
membership ties is a political process: the inclu- ganization (NGO) stakeholders. Through these
sion (and exclusion) of particular organizations processes, participants constructed themselves
can be highly contentious and can significantly as closely and directly tied to each other. The
affect collaborative outcomes (Gray, 1985; Hardy relationship between particularized member-
& Phillips, 1998). Participants with power may ship ties and collective identity leads to our next
use it to exclude other stakeholders (Gricar & proposition.
Brown, 1981) in order to avoid “extreme” or “rad-
ical” views (Warren, Rose, & Bergunder, 1974). In Proposition 3: Participants in an inter-
addition, who has authority, which procedures organizational collaboration will be
are invoked, and what particular coalitions are more likely to discursively construct a
formed all have political implications, and, as collective identity when they engage
such, participants may use power to influence in conversations that establish partic-
particularized membership ties. At the same ularized membership ties among
time, the larger discursive context constrains them.
action; to be able to participate in a collabora-
tion, participants need to occupy some legiti-
CONVERSATIONS IN COLLABORATION
mate position (Gray, 1985), but within any dis-
course, only a limited number of subject The discursive construction of a collective
positions are understood as meaningful, legiti- identity may make collective action possible,
mate, and powerful (Fairclough 1992; Foucault, but it does not necessarily lead to effective col-
1972). Hence, inclusion or exclusion of certain laboration (cooperative, interorganizational ac-
groups, by virtue of whether they are legitimate tion that produces innovative, synergistic solu-
participants in a collaboration or not, may tions and balances divergent stakeholder
largely be determined by broader discourses concerns). We argue that collective identity is
(Hardy & Phillips, 1999). translated into effective collaboration through
The formation of particularized membership the impact of two other aspects of conversations.
ties through conversation is an important pre- First is “constructions of key issues,” by which
cursor to the discursive production of a collec- we mean the ways in which central issues are
tive identity in the context of collaboration. defined in conversation. Second is “styles of
66 Academy of Management Review January

talk” that characterize these conversations— lished in participants’ conversations: (1) com-
their tone, style, rhythm, and format. Each of mon constructions of key issues and (2) private
these aspects of conversation makes important constructions of those same issues. As we dis-
contributions to the process, but most critical to cussed with generalized and particularized
producing effective collaboration are the ten- membership ties, these constructions may be
sions, in the form of conflict and ambiguity, produced in the same, overlapping, or different
among different constructions and styles of talk. conversations. We treat them separately for the-
oretical clarity, but any empirical investigation
would need to be attuned to this potential over-
Constructions of Key Issues
lap.
In collaborations, conversations must produce Common constructions. In order for effective
resources on which participants can draw to collaboration to occur, participants must engage
make sense of the situation and to create and in conversations that produce resources to make
legitimate courses of action. These discursive sense of the issue and related matters in terms
resources— understandings, frames of refer- on which there is general agreement. We call
ence, and forms of knowledge, or, as we call these discursive objects “common construc-
them, “constructions”—are similar to Swidler’s tions.” They occur when participants negotiate a
conception of culture as a “‘tool kit’ . . . from general agreement regarding the causes, symp-
which actors select differing pieces for con- toms, assumptions, and potential solutions that
structing lines of action” (1986: 277). Construc- relate to the issue around which the collabora-
tions provide a discursive tool kit from which tion is formed. These constructions are essen-
participants can select and, having selected tial, because they allow participants to make
some constructions over others, engage in cer- sense of complex problems (Weick & Roberts,
tain actions over others. A construction “rules 1993) and to communicate with each other well
in” certain ways of thinking about, talking enough to coordinate collective action (Huxham,
about, and acting on a topic and “rules out” 1996). It is not enough to be connected to other
other ways (Hall, 2001: 72). In the process of col- participants and to the issue; participants must
laboration, this occurs as the issue that has also establish a basis for communication if they
brought the collaborating partners together is are to reconcile their different interests and if
described, examined, analyzed, and acted on; synergy is to occur (Gray, 1989; Trist, 1983).
these conversations discursively produce and The production of common constructions is
maintain certain forms of knowledge, frames of distinct from the establishment of the general-
reference, and understandings. ized membership ties that help to produce col-
Our perspective on the construction of key is- lective identity. Participants may engage in con-
sues is distinct from one in which the under- versations that recognize their common interest
standing of a problem is cognitive in nature; a in an issue without necessarily constructing it in
cognitive perspective might focus on “informa- the same terms. For example, union representa-
tion” and the ways in which it is collected, tives and managers involved in wage negotia-
shared, processed, and analyzed in order to de- tions may well produce generalized ties that
termine the correct course of action (Huber, connect them to a successful business. Through
1991). In contrast, from a discursive perspective, those negotiations they may even develop
through conversations associated with the col- strong particularized ties that connect them to
laboration, individuals construct and legitimate each other as negotiators. Accordingly, they
particular knowledge, frames of reference, and may develop a collective identity around their
understandings, which, in turn, make certain ac- joint roles in industrial bargaining. But they will
tions meaningful and appropriate. Thus, the not necessarily have a common construction of
critical significance of constructions is their the problem: managers may construct wage in-
availability to participants to use as resources creases as a threat to competitiveness, while
in further conversations (e.g., securing legiti- union representatives may construct them as a
macy and commitment for particular courses of redistribution of profits. In such a situation, it
action). becomes difficult to produce a synergistic out-
We argue that for effective collaboration to come; instead, the “solution” is more likely to be
occur, two types of constructions must be estab- imposed, if one party proves more powerful, or
2005 Hardy, Lawrence, and Grant 67

to be a compromise, if not. While this form of novation arises (Jehn, 1997; Lovelace, Shapiro, &
action may be common in negotiations (Wells & Weingart, 2001). In other words, while common
Liebman, 1996), it does not constitute effective constructions facilitate communication and al-
collaboration. low participants to create an atmosphere of mu-
Common constructions are necessary for ef- tual understanding, private constructions allow
fective collaboration, because communication creativity and innovation (cf. Dougherty, 1996;
among participants is contradictory and confus- Fiol, 1995). The value of diversity to innovation is
ing without them. For example, in a study of a well-established idea in the collaboration lit-
interdepartmental collaboration associated erature, which often recommends bringing to-
with innovation, Dougherty (1996) found that dif- gether individuals with different backgrounds
ferent departments employed different mean- to generate new ideas (e.g., Gray, 1989; Huxham,
ings of the term market-oriented innovation. For 1996), We suggest, however, that it is not the
R&D, it meant product specifications and tech- diversity of the collaborators per se that is the
nical features—what the product could do. In crucial element but, rather, the production of
manufacturing, it meant durability and reliabil- constructions of the issue that are clearly at-
ity, which translated into fewer features and tached to particular participants.
simpler specifications. Members in the planning The articulation of private constructions pro-
department thought it meant having the product vides a set of discursive resources on which
in the right market niche. Although these de- participants can draw as they attempt to nego-
partments got together to collaborate and saw tiate solutions to the issue at hand. As such,
themselves as connected to the same issue, par- private constructions play an important role in
ticipants failed to produce a common construc- ensuring that the actions that emerge out of
tion of this issue. As a result, attempts to inno- collaboration balance stakeholder interests, be-
vate failed, because the would-be innovators cause they publicly commit participants to take
failed to bridge their different “thought worlds” into account the interests of the organizations
through the articulation of common construc- they represent (Salancik, 1977) and because they
tions, even though they engaged in conversa- help to ensure that the substantive issues pre-
tions with each other and used the same termi- occupying the organizations represented in the
nology (Dougherty, 1996). collaboration are present in the conversations
Private constructions. Effective collaboration among participants. In fact, private construc-
also requires participants to engage in conver- tions are likely to be rooted in conversations
sations that produce private constructions. In that occur in participants’ home organizations:
using this term, we are not referring to undis- as participants discuss the progress of the col-
closed constructions held by participants but, laboration with their managers, private con-
rather, constructions that are produced in ways structions will be produced, which then can be
that attach them to particular participants in- articulated in conversations among collabora-
stead of the group as a whole. Thus, private tors.
constructions are discursive objects that lead Effective collaboration is more likely to ensue
individuals to make sense of and express key if conversations produce both common and pri-
issues in disparate and often conflicting terms. vate constructions on an ongoing basis. Com-
In the context of interorganizational collabora- mon constructions provide a basis for moving
tion, they are often associated with the goals of forward; private constructions facilitate the cre-
the different organizations represented in the ation of the task-oriented conflict among partic-
collaboration, but they might also involve a ipants necessary for innovation (Cosier & Rose,
wide range of issues, such as appropriate tech- 1977; Jehn, 1997). Private constructions are also
nologies to employ, time frames for the collabo- an important way to ensure that stakeholder
ration, and stakeholder involvement. interests are not neglected through both public
Private constructions play a critical role in commitments and the inclusion of substantive
fostering creativity by juxtaposing “widely di- concerns.
vergent bodies of knowledge and experience” We argue that both sets of constructions need
(Fiol, 1995: 71), whereas conversations that in- to be produced on an ongoing, rather than se-
volve only stable, shared knowledge inhibit the quential, basis; this argument contrasts with the
discovery of contrary experience from which in- group development literature, which tends to
68 Academy of Management Review January

suggest a period of divergence, such as brain- resources available to all participants will be
storming, followed by convergence prior to ac- and the greater the motivation will be for par-
tion (Katzenbach & Smith, 1993; Mohrman, Co- ticipants to ensure that the tension between
hen, & Mohrman, 1995). Our argument for the common and private constructions is main-
ongoing juxtaposition of common and private tained. This is not to say that, as the collabora-
constructions stems from the perspective that tion develops, there will be not particular mo-
language does not represent or reveal the facts ments where convergence occurs, allowing
of the world but constructs them in conversation; decisions to be made and coordinated action to
thus, facts are not established once and for all be taken. However, thereafter, new issues will
but must continuously be reproduced if they are arise, and tension must again be created and
to be meaningful conversational elements that maintained until convergence once again oc-
can affect the actions ensuing from the collabo- curs.
ration. The second factor stems from the impact of the
To summarize, effective collaboration is facil- composition of participants on the dynamics of
itated by a tension that stems from the ongoing conversations (Westley, 1990). We argue that the
circulation and interplay of common and private introduction of new members, in terms of either
constructions and the ambiguities that result. participating organizations or individual repre-
While representational perspectives of lan- sentatives, will be associated with both the in-
guage and communication “view ambiguity as troduction of new private constructions and the
a problem to be solved,” a discursive approach examination of the differences between common
“treats ambiguity as an opportunity to chal- and private constructions; the conversations
lenge, skirt, and reinvent received knowledge” prompted by the arrival of new participants will,
(Eisenberg, 1998: 97). Thus, the innovation asso- therefore, offset participants’ tendencies toward
ciated with effective collaboration is fueled by convergence.
the ambiguity that emerges out of the simulta- Finally, formal barriers to convergent lan-
neous production of common and private con- guage can be erected in the institution of collab-
structions. These dynamics lead to our next orative practices, such as situations where par-
proposition. ticipants are required to articulate private
constructions for the purposes of creating con-
Proposition 4: The relationship be-
flict. These are not the only factors that influ-
tween collective identity and effective
ence the production of common and private con-
collaboration will be strengthened by
structions, but they represent three important
the maintenance of an ongoing ten-
factors, as indicated below.
sion between common and private
constructions of key issues throughout Proposition 5: The ongoing production
the life of the collaboration. of common and private constructions
Sustaining the production of both common in conversations will be facilitated by
and private constructions of key issues over the (a) mutual awareness of conversations
life of a collaboration is made difficult by peo- across the collaboration and the rep-
ple’s preferences for coherence in conversation resentatives’ home organizations, (b)
(Woodilla, 1998). A number of factors can miti- the introduction of new members over
gate this tendency toward convergence, how- time, and (c) barriers to convergence
ever, and can facilitate the ongoing production in the collaboration.
of both sets of constructions. The first factor
stems from the relationship between conversa-
Styles of Talk
tions that occur inside and those that occur out-
side the collaboration. We argue that the greater The second feature of conversations that mod-
the awareness of the content of those conversa- erates the relationship between collective iden-
tions on both sides (participants are aware of tity and the likelihood of effective collaboration
other participants’ conversations in their home is the style of talk in which actors engage—the
organizations, and home organization manag- patterns in tone, style, rhythm, and format of
ers are aware of the conversations within the conversations. Styles of talk are critical in col-
collaboration), the more diverse the discursive laboration because they provide the emotional
2005 Hardy, Lawrence, and Grant 69

energy necessary for participants to translate a In collaboration, problem-solving decisions are

collective identity into effective collaboration. eventually taken by a group of stakeholders who
have mutually authorized each other to reach a
The lack of market and hierarchical mecha- decision. Thus, power to define the problem and
nisms of control among participants inherent in to propose a solution is effectively shared among
collaboration means that participants must gen- the decision makers (Gray, 1989: 119).
erate their own motivation if they are to collab-
orate effectively. Different styles of talk gener- Thus, through cooperative talk, action is gen-
ate this motivation in different ways and with erated collectively and inclusively. The adop-
different results. As with our discussion of the tion of such a style of talk ensures that commu-
construction of key issues, it is not the utilization nicative activity is shared and that meaning is
of a single style or even a combination of styles coproduced by all participants, including less
that is critical to the accomplishment of effective powerful partners. As Warren et al. point out,
collaboration but, rather, the creation and main- simply including diverse stakeholders is not
tenance of a tension among different styles of enough to ensure effective collaboration, since
talk by participants. there is always the risk that their voices will
We argue that effective collaboration involves simply be “outweighed and outheard” (1974:
an ongoing tension between two styles of talk in 127). Cooperative talk works to ensure that the
particular: (1) cooperative talk, which arises views of weaker organizational partners are in-
when the tone, style, rhythm, and format of con- corporated, thereby enabling some form of
versations emphasize the willingness of partic- shared purpose to emerge. It also helps partici-
ipants to listen to and engage with each other’s pants to maintain their common constructions
as central to the conversation and to integrate
positions and interests, and (2) assertive talk,
private constructions in a way that is more
which occurs when the character of conversa-
likely to produce the synergistic outcomes asso-
tions emphasizes participants’ insistence on ar-
ciated with effective collaboration (e.g., Lewis &
ticulating their own views and positions. In the
Weigert, 1985; Sabel, 1993). Moreover, coopera-
remainder of this section, we explore the ways
tive talk tends to reinforce a group’s collective
in which these two styles of talk influence the
identity—in signaling a willingness to cooper-
relationship between collective identity and the
ate, participants will draw on the collective
likelihood of effective collaboration, arguing
identity as a discursive resource, talking in
that both forms are necessary for effective col-
terms of “we” and “us,” and recreating and re-
laboration to occur. vising identity in an ongoing manner.
Cooperative talk. Cooperative talk tends to A study of a collaboration between an NGO
occur in conversations that emphasize partici- and a number of refugee organizations in the
pants’ similarity, mutual affiliations, and United Kingdom (Hardy & Phillips, 1998) pro-
shared interests. This emphasis might be cre- vides an example of the importance of coopera-
ated through particular words chosen (or tive talk among differentially powerful partners.
avoided), a tone that suggests mutual respect This collaboration involved a considerable em-
and consideration, or a structure in which all phasis on cooperative talk, particularly on the
participants play an equal role (Atkinson & Her- part of the more powerful NGO, whose represen-
itage, 1984; Psathas, 1995; Woodilla, 1998). By tatives were constantly questioning, “Should we
conducting conversations in this manner, partic- be doing this or should we be helping a commu-
ipants associate the content of those conversa- nity organization to do it? You have to introduce
tions with a willingness to work with and accom- that questioning so you don’t automatically as-
modate each other’s interests and preferences. sume that you’ve got to do it” (collaborator
This can occur even when the topic of conversa- quoted in Hardy & Phillips, 1998: 222). Represen-
tion revolves around participants’ private con- tatives of the NGO set great store on full partic-
structions and particular interests; through the ipation by refugees as equal partners: “We don’t
tone, style, and format of their conversation, par- usually impose things on [these] groups, we try
ticipants can signal accommodation and a will- to facilitate it” (collaborator quoted in Hardy &
ingness to compromise. Such cooperative talk Phillips, 1998: 222). Hardy and Phillips concluded
underlies many normative models of collabora- that this collaboration was effective because of
tion: the way in which all partners— but particularly
70 Academy of Management Review January

the more powerful participants—voluntarily are committed to organizational interests. As a

and cooperatively participated in it, compared result, they will be more likely to continue their
to other relationships that appeared to be col- support for collaboration, which, in turn, will
laborative on the surface but where neither par- motivate participants to contribute to the con-
ticipation nor power was shared. versations instead of the reverse—wondering
Assertive talk. Assertive talk aids in effective whether their organization is likely to pull out of
collaboration by helping participants ensure the collaboration, which makes it difficult for
that their organizational interests are repre- participants to sustain energy (cf. Westley, 1990).
sented in any action generated by the collabo- The danger that stems from lack of assertive
ration. Assertive talk energizes the differences talk is illustrated in Westley and Vredenburg’s
among participants and provides a motivational (1991) study of a collaboration between a super-
basis for achieving synergistic, rather than sim- market chain and an environmental organiza-
ply compromising, outcomes. By engaging in as- tion in Canada to develop a new line of environ-
sertive talk, individual participants attempt to mentally friendly products. The two CEOs
influence the direction of conversations and engaged in highly cooperative talk and sup-
frame the discussions that go on within them ported a common construction of the chal-
(Westley, 1990). As with cooperative talk, asser- lenge—the importance of environmental groups
tive talk can be produced through a conversa- and business working together. This coopera-
tion’s tone (highly pitched, self-assured, insis- tive talk was not, however, complemented with
tent), style (aggressive, uncompromising), or assertive talk, since the CEO of the environmen-
structure (forceful turn-taking, free-for-all). It al- tal organization attempted to “streamline” meet-
lows individual participants to discursively pro- ings and to constrain the opportunity for other
duce positions in which they are visibly and participants in the collaboration to engage in
audibly affected by their personal and organi- assertive talk. This made it difficult for dissent-
zational stakes in the issue at hand. It also en- ing employees to promote or defend private
sures that private constructions are articulated constructions of the issues, and it reduced the
strongly and can be explored, promoted, and degree to which the interests of outside stake-
pursued; without assertive talk, private con- holders were considered. As a result, the collab-
structions are more likely to remain peripher- oration failed to produce innovative outcomes,
al—stated, but not pushed forward for inclusion lost the support of the board of the environmen-
in collective sensemaking and decision-making tal organization, and was publicly condemned
processes. Moreover, assertive talk can be im- by other high-profile environmental groups.
portant in holding up common constructions to We argue that a combination of cooperative
scrutiny, whereas cooperative talk tends to rein- and assertive talk will have a positively mod-
force them. Since effective collaboration de- erating effect on the relationship between col-
pends on innovation and creativity, the forceful lective identity and effective collaboration. Co-
promotion of private constructions and the crit- operative talk facilitates the leveraging of
ical examination of common constructions are common constructions and the integration of
both essential elements in this process. private constructions, both of which engender
Assertive talk is also important for the way in collective action. Assertive talk ensures that pri-
which it constructs participants’ representa- vate constructions are advanced, organizational
tional roles as legitimate and central aspects of responsibilities are acknowledged, and com-
the process. In engaging in assertive talk, par- mon constructions are critically challenged, all
ticipants demonstrate their commitment to the of which help to produce innovation and ensure
proprietary interests of the organizations they that stakeholder interests are represented. Most
represent. This can smooth what has been de- critically, effective collaboration is driven by an
scribed in collaboration research as the “two- ongoing interplay of both styles of talk and will
table problem” (Gray, 1989), where participants emerge from situations in which agreement is
are perceived as overly identified with the col- regularly challenged and where styles of talk
laboration rather than the organization in which vary over time and across conversations in a
they work. Consequently, assertive talk can in- fluid and potentially unpredictable manner.
crease the degree to which home organization Each style of talk plays a critical role in the
managers perceive that their representatives accomplishment of effective collaboration, and
2005 Hardy, Lawrence, and Grant 71

neither can be suspended or set aside. They to change the dominant style of talk and switch
need not always feature together in an individ- to the alternative form.
ual conversation, but they will have the greatest
Proposition 7: The ongoing production
impact if both are interwoven over the life of the
of cooperative and assertive styles of
collaboration. In summary, a tension between
talk will be facilitated by (a) legitima-
the styles of talk helps to promote innovation
tion of both styles of talk within the
and synergy and to ensure that stakeholder in-
collaboration, (b) heterogeneity of
terests are represented. These dynamics lead to
roles among participants, and (c) tem-
our next proposition.
poral discontinuities in conversation.
Proposition 6: The relationship be-
The penultimate point to make about our
tween collective identity and effective
model concerns the various feedback loops that
collaboration will be strengthened by
exist among the concepts we have discussed.
the maintenance of an ongoing ten-
First, the relationship between generalized and
sion between assertive and coopera-
particularized ties and common constructions is
tive styles of talk throughout the life of
complex. Common constructions are distinct
the collaboration.
from membership ties, although both can be re-
The ongoing production of both cooperative sources for the production of common construc-
and assertive talk may be difficult to achieve tions: collaborators can draw on the connections
because of the potentially self-reinforcing as- to the issue and to each other as they elaborate
pects of each style of talk: norms of reciprocity the way in which they jointly construct the issue.
suggest that cooperative talk on the part of one At the same time, common constructions act as
participant is likely to engender cooperative resources for the ongoing discursive accom-
talk by others; similarly, assertive talk is likely plishment of these ties, and, in fact, both ties
to be met with equally assertive talk (Andersson and common constructions may be produced in
& Pearson, 1999). Thus, factors that increase the the same conversations.
ongoing production of both styles of talk will be Second, styles of talk are also linked to mem-
those that mitigate these self-reinforcing ten- bership ties in a reciprocal relationship. Gener-
dencies. alized and particularized ties are resources for
One such factor is the legitimacy of both both styles of talk, depending on the emotional
styles of talk within the collaboration. This reaction to the issue and to the partners. For
might be affected by direct discussions of the example, a passion for the issue (generalized)
issue within the collaboration (as might be and respect for partners (particularized) are
prompted by a skilled facilitator) or by the likely to produce cooperative talk, whereas a
broader discursive context surrounding the col- passion for the issue and a suspicion of partners
laboration (with different sectors and industries may produce assertive talk. The style of talk
having distinct norms that might challenge or also influences ties. For example, the predomi-
privilege one style of talk or another). A second nance of cooperative talk may weaken general-
factor concerns the diversity of roles or identi- ized ties, but if, as a result, the participative
ties that participants take on. From a discursive process becomes more important than the end
perspective, different identities are associated itself, particularized ties will strengthen. In con-
with different legitimated vocabularies and con- trast, the predominance of assertive talk may
versational styles (Fowler, 1991; Iedema, 1998). strengthen generalized ties but weaken partic-
By ensuring a heterogeneity of roles for partici- ularized ties.
pants within the collaboration, a single domi- Finally, and more generally, to sustain effec-
nant style of talk is less likely to emerge. Fi- tive collaboration, the processes underpinning
nally, collaborative practices such as the pacing the model must remain ongoing and success-
of conversations can help sustain both coopera- ful—hence, the feedback loop shown between
tive and assertive talk. Styles of talk are more “effective interorganizational collaboration”
likely to be self-reinforcing within the context of and “generalized” and “particularized member-
continuous conversations; consequently, the ship ties” in Figure 1. Any deviation or break-
more often conversations are broken up, the down of these intricate processes would lead to
greater the opportunity will be for participants the failure of the collaboration. Thus, we are
72 Academy of Management Review January

suggesting in this model that the process of sions among them. Our discursive perspective
achieving effective collaboration may be a rel- also highlights the role of broader discourses in
atively fragile one. supporting and constraining conversations; fu-
ture research might examine the impacts and
dynamics of the particular discourses within
CONCLUSION
which an interorganizational collaboration is
We have developed a theory of effective col- enacted.
laboration that focuses on the roles of collective A second research implication concerns the role
identity and conversation. Having defined effec- of stakeholders. Collaboration researchers have
tive collaboration as cooperative, interorganiza- often adopted a social psychological perspec-
tional action that produces innovative, synergistic tive, focusing on group dynamics and the behav-
solutions and balances divergent stakeholder ior of individuals within the collaboration, while
concerns, we adopted a discourse-oriented per- giving the interactions between stakeholder
spective that highlights the interaction of the dis- and the collaboration rather less attention. Our
cursive context, conversations among partici- model suggests that discourse-oriented studies
pants, and the discursive resources that are can make an important contribution to our un-
critical in achieving effective collaboration. We derstanding of this relationship when the con-
have argued that the discursive production of a versations between organizational members
collective identity is an important step in produc- and their representatives on the collaboration
ing effective collaboration and that it depends on are examined. Researchers could examine col-
the production of generalized and particularized laborative failures caused by the withdrawal of
membership ties. We have further argued that the organizational support and identify the signals
relationship between collective identity and effec- to which organizations react when assessing
tive collaboration will be stronger to the extent whether to continue their support. It is not clear
that conversations among participants sustain an whether organizational members respond more
ongoing tension between common and private to formal texts, such as progress reports, or to
constructions and between cooperative and asser- informal conversations with representatives.
tive styles of talk. This theory has important im- Similarly, researchers could examine the sym-
plications for research on collaboration and for bols that reassure stakeholders that a collabo-
the study of language in organizations more gen- ration is safe and productive and, consequently,
erally, as well as significant practical implica- worthy of continued support.
tions. A third implication concerns the factors that
facilitate the maintenance of the tensions be-
tween common and private constructions and
Implications for Research
between cooperative and assertive styles of
The first implication is for research on the role talk. Although the tradeoff between partici-
of language in effective collaboration. Tradi- pants’ roles as collaborators and as stakeholder
tionally, studies of interorganizational collabo- representatives has long been recognized in the
ration have tended to privilege action at the collaboration literature (Gray, 1985, 1989), it is
expense of talk; effective collaboration has been less clear how such a tradeoff can be managed
associated with “getting things done” while effectively. We have identified specific ways of
avoiding “talk-fests.” We believe in the impor- maintaining the tensions that produce effective
tance of action, but we focus on the way in collaboration. Thus, in future research scholars
which talk establishes the foundations for ac- should pay close attention to such factors as
tion in effective collaboration. barriers to convergence, introduction of new
Our model suggests some specific directions members, and transparency of conversations in
for research on this relationship. Researchers stakeholder organizations (which affect the like-
interested in the role of language should focus lihood that a tension between common and pri-
on the discursive resources (generalized and vate constructions will be maintained), as well
particularized membership ties, collective iden- as temporal discontinuities in the process, het-
tity, common and private constructions of key erogeneity of participants, and the legitimacy of
issues, and cooperative and assertive styles of various styles of talk (which affect the partici-
talk) that are available and, especially, the ten- pants’ styles of talk).
2005 Hardy, Lawrence, and Grant 73

A fourth research implication stems from the might attend to a broader range of organiza-
methodological advantages associated with the tional issues by considering the tensions ef-
adoption of a discursive perspective. Studies of fected by the presence of alternative construc-
discourse are strongly empirical because of the tions and styles of talk. We believe that much is
way in which discourse leaves “traces” (Taylor to be gained by developing a wider array of
& Van Every, 2000), meaning that the objects of methods that can highlight primary patterns of
analysis are observable instances of talk and interaction without submerging less dominant
text. Accordingly, proponents of a discursive ap- threads.
proach view the social world as an ongoing ac-
complishment that is held precariously in place
Implications for Practice
through discursive struggle, which allows us to
view a collaboration as something that is al- Our model has important practical implica-
ways in the act of “becoming” (cf. Tsoukas & tions for organizational actors attempting to de-
Chia, 2002), rather than as a discrete entity. A velop effective collaborations. Participants first
discursive approach also allows us to examine need to produce a collective identity through the
the “levels” involved in collaboration without establishment of both generalized and particu-
reifying them. Whereas traditional research has larized membership ties. Practically speaking,
focused on the psychology of individual partic- the opening sequences of collaboration should
ipants, the group dynamics of the collaboration involve conversations that connect the issue to
team, or, less often, the stakeholders repre- potential participants and articulate specific
sented in the collaboration, a discursive ap- connections between them. Generalized mem-
proach allows us to conceptualize these interac- bership ties rely on the establishment and legit-
tions through conversations that embrace and imation of a vocabulary from which participants
interconnect all “levels.” can draw to explain their own and others’ par-
A final implication is connected to the emerg- ticipation to each other and to their home orga-
ing focus on organizing, rather than organiza- nizations; particularized ties require discursive
tion (Weick, 1998), and the conceptualization of resources that can be used to describe and ex-
organizational change as a process of becoming plain the relationships among participants.
(Tsoukas & Chia, 2002). This perspective high- Once a collective identity is discursively pro-
lights the importance of developing theories duced, collaborators need to shift the conversa-
that attend to the dynamic, ongoing nature of tions toward the ongoing production of common
organizational processes. Our model has shown and private constructions. Our analysis sug-
how the maintenance of ongoing tensions with gests that this can be facilitated by establishing
respect to the constructions produced in conver- formal procedures in the collaboration that fo-
sation and styles of talk is connected to effective cus on repeatedly articulating the concerns of
collaboration through the way in which these managers in the home organization; by building
tensions balance the competing needs of partic- structures into the home organizations, such as
ipant and representative in an ongoing process. teams, whose focus is integrating the learning
Similar tradeoffs exist in other organizational occurring in the collaboration; by rotating mem-
processes (e.g., interdepartmental collaboration, bership in terms of who represents the home
cross-functional project teams, or taskforces organization and, where appropriate, by rotat-
where the individual is both a member of a ing or adding to the stakeholders represented in
project team or taskforce and a representative of the collaboration; and by including in the col-
a department or unit). In fact, organizational laboration process some formal joint analysis of
membership generally is fraught with conflict- the differences among participants’ private con-
ing goals and the need to balance individualis- structions and between common and private
tic and collectivist inputs. constructions. These mechanisms help to main-
In line with this perspective, our model explic- tain the salience of both common and private
itly focuses on how discourse theory enables us constructions, as well as help to maintain the
to observe the means by which individuals distinction between them. More imaginative
maintain multiple conflicting constructions of mechanisms might include role plays in which
key issues and styles of talk on an ongoing participants swap representational roles for a
basis. In doing so, it shows how researchers period of time and are encouraged to voice
74 Academy of Management Review January

“strong” descriptions of others’ private construc- when the stakes are high, but when they are,
tions. managers and researchers would do well to lis-
Facilitating an ongoing tension between as- ten to the talk of collaboration.
sertive and cooperative styles of talk over the life
of a collaboration requires an array of different
skills, structures, and processes (Schweiger,
REFERENCES
Sandberg, & Ragan, 1986). At a minimum, the
importance of this tension suggests the value of Abrams, D., & Hogg, M. A. (Eds.). 1999. Social identity and
social cognition. Oxford: Blackwell.
skilled facilitation that can engender coopera-
tive styles of talk amidst conflict and can engen- Albert, S., & Whetten, D. 1985. Organizational identity. Re-
search in Organizational Behavior, 7: 263–295.
der legitimate assertive talk despite a group’s
desire to “get along.” The importance of this Altheide, D. L. 1988. Mediating cutbacks in human services:
A case study in the negotiated order. Sociological Quar-
tension also suggests the need to legitimate
terly, 29: 339 –355.
both forms of talk among participants and,
Alvesson, M., & Kärreman, D. 2000. Varieties of discourse: On
where possible, to embed that legitimacy in con-
the study of organizations through discourse analysis.
versations outside as well as inside the collab- Human Relations, 53: 1125–1149.
oration. In addition—and given that different
Andersson, L., & Pearson, C. 1999. Tit for tat? The spiraling
identities are associated with different legiti- effect of incivility in the workplace. Academy of Man-
mated vocabularies and conversational styles— agement Review, 24: 452– 471.
ensuring heterogeneous roles are assigned to Ashforth, B. E., & Mael, F. 1989. Social identity theory and the
and rotated among participants may help to pre- organization. Academy of Management Review, 14: 20 –
vent one form of talk from dominating. Finally, 39.
more use might be made of temporal disconti- Astley, W. G. 1984. Toward an appreciation of collective
nuities to engender a change in the style of talk, strategy. Academy of Management Review, 9: 526 –535.
by a facilitator calling “time out” or by the de- Atkinson, J. M., & Heritage, J. (Eds.). 1984. Structures of social
liberate pacing of the meeting schedule. action: Studies in conversation analysis. Cambridge:
A final practical implication concerns the Cambridge University Press.
boundaries of when the model can and should Benford, R. D., & Snow, D. A. 2000. Framing processes and
be applied. Our intent has been to develop a social movements: An overview and assessment. An-
nual Review of Sociology, 26: 611– 639.
theory that explains the role of discourse in pro-
ducing effective collaboration. If organizational Bresser, R. K., & Harl, J. E. 1986. Collective strategy: Vice or
virtue? Academy of Management Review, 11: 408 – 427.
actors engage in the processes described here,
they should increase their chances of achieving Cerulo, K. A. 1997. Identity construction: New issues, new
directions. American Review of Sociology, 23: 385– 409.
cooperation and innovation while also meeting
their respective interests. However, not all orga- Chalaby, J. K. 1996. Beyond the prison-house of language:
Discourse as a sociological concept. British Journal of
nizations, and not all forms of collaboration, are
Sociology, 47: 684 – 698.
so ambitious. A significant proportion of interor-
Collins, R. 1981. On the microfoundations of macrosociology.
ganizational relationships are not intended to
American Journal of Sociology, 86: 984 –1013.
achieve innovative solutions or balance stake-
Cosier, R., & Rose, G. 1977. Cognitive conflict and goal con-
holder interests; organizational actors regularly
flict effects on task performance. Organizational Behav-
enter into relationships that are aimed at nego- ior and Human Performance, 19: 378 –391.
tiating stakes in some fixed pot of resources,
Deetz, S. A. 1992. Democracy in an age of corporate coloni-
rapidly solving some temporary problem, or zation. Albany: State University of New York Press.
simply gaining legitimacy in the eyes of a third
Dougherty, D. 1996. Organizing for innovation. In S. R. Clegg,
party. In none of these cases would we suggest C. Hardy, & W. R. Nord (Eds.), Handbook of organization
that actors engage in the processes described studies: 424 – 439. London: Sage.
above. Our model shows how large an invest- Dutton, J. E., & Ashford, S. J. 1993. Selling issues to top man-
ment and commitment effective collaboration agement. Academy of Management Review, 18: 387– 428.
demands in terms of the ambiguity that partici- Dutton, J. E., Dukerich, J. M., & Harquail, C. V. 1994. Organi-
pants and stakeholders have to tolerate, as well zational images and member identification. Administra-
as the need for subsidiary sets of supporting tive Science Quarterly, 39: 239 –263.
conversations to be set up in home organiza- Dyer, J., & Singh, H. 1998. The relational view: Cooperative
tions. Such an investment is only worthwhile strategies and sources of interorganizational competi-
2005 Hardy, Lawrence, and Grant 75

tive advantage. Academy of Management Review, 23: Hall, S. 2001. Foucault: Power, knowledge and discourse. In
660 – 679. M. Wetherell, S. Yates, & S. Taylor (Eds.), Discourse the-
ory and practice: A reader: 72– 81. London: Sage.
Eisenberg, E. M. 1998. Flirting with meaning. Journal of Lan-
guage and Social Psychology, 17: 97–108. Hardy, C., Lawrence, T. B., & Phillips, N. 1998. Talk and
action: Conversations and narrative in interorganiza-
Elsbach, K. D., & Kramer, R. M. 1996. Members’ responses to
tional collaboration. In D. Grant, T. Keenoy, & C. Oswick
organizational identity threats: Encountering and coun-
(Eds.), Discourse and organization: 65– 83. London: Sage.
tering the Business Week ratings. Administrative Sci-
ence Quarterly, 41: 442– 476. Hardy, C., & Phillips, N. 1998. Strategies of engagement:
Lessons from the critical examination of collaboration
Fairclough, N. 1992. Discourse and social change. Cam-
and conflict in an interorganizational domain. Organi-
bridge: Polity Press.
zation Science, 9: 217–230.
Fiol, C. M. 1991. Managing culture as a competitive resource:
Hardy, C., & Phillips, N. 1999. No joking matter: Discursive
An identity-based view of sustainable competitive ad-
struggle in the Canadian refugee system. Organization
vantage. Journal of Management, 17: 191–211.
Studies, 20: 1–24.
Fiol, C. M. 1995. Thought worlds colliding: The role of con-
Heide, J. 1994. Interorganizational governance in marketing
tradiction in corporate innovation processes. Entrepre-
channels. Journal of Marketing, 50: 40 –51.
neurship Theory and Practice, 19: 71–90.
Heracleous, L., & Barrett, M. 2001. Organizational change as
Fiol, C. M. 2002. Capitalizing on paradox: The role of lan-
discourse: Communicative actions and deep structures
guage in transforming organizational identities. Orga-
in the context of information technology implementa-
nization Science, 13: 653– 666.
tion. Academy of Management Journal, 44: 755–778.
Ford, J. D., & Ford, L. W. 1995. The role of conversations in
Hogg, M. A. 2000. Social identity and self-categorization pro-
producing intentional change in organizations. Acad-
cesses in organizational contexts. Academy of Manage-
emy of Management Review, 20: 541–570.
ment Review, 25: 121–140.
Foucault, M. 1972. The archeology of knowledge. London:
Huber, G. 1991. Organizational learning: The contributing
Routledge.
process and the literatures. Organization Science, 2: 88 –
Fowler, R. 1991. Language in the news: Discourse and ideol- 115.
ogy in the press. London: Routledge.
Huxham, C. (Ed.). 1996. Creating collaborative advantage.
Gheradi, S. 1995. Gender, symbolism and organizational cul- London: Sage.
tures. London: Sage.
Iedema, R. A. M. 1998. Institutional responsibility and hidden
Gioia, D. A., Schultz, M., & Corley, K. G. 2000. Organizational meanings. Discourse and Society, 9: 481–500.
identity, image, and instability. Academy of Manage-
Jehn, K. A. 1997. A qualitative analysis of conflict types and
ment Review, 25: 63– 81.
dimensions in organizational groups. Administrative
Gioia, D. A., & Thomas, J. B. 1996. Identity, image, and issue Science Quarterly, 42: 530 –557.
interpretation: Sensemaking during strategic change in
Katzenbach, J. R., & Smith, D. K. 1993. The wisdom of teams:
academia. Administrative Science Quarterly, 41: 370 –
Creating the high performance organization. Boston:
403.
Harvard Business School Press.
Grant, D., Keenoy, T., & Oswick, C. 1998. Introduction: Orga-
Knights, D., & Morgan, G. 1991. Strategic discourse and sub-
nizational discourse: Of diversity, dichotomy and multi-
jectivity: Towards a critical analysis of corporate strat-
disciplinarity. In D. Grant, T. Keenoy, & C. Oswick (Eds.),
egy in organizations. Organization Studies, 12: 251–273.
Discourse and organization: 1–14. London: Sage.
Lawrence, T. B., Hardy, C., & Phillips, N. 2002. Institutional
Gray, B. 1985. Conditions facilitating interorganizational col-
effects of interorganizational collaboration: The emer-
laboration. Human Relations, 38: 911–936.
gence of proto-institutions. Academy of Management
Gray, B. 1989. Collaborating: Finding common ground for Journal, 45: 281–290.
multiparty problems. San Francisco: Jossey-Bass.
Lawrence, T. B., Phillips, N., & Hardy, C. 1999. Watching
Gray, B., & Hay, T. M. 1986. Political limits to interorganiza- whale-watching: A relational theory of organizational
tional consensus and change. Journal of Applied Behav- collaboration. Journal of Applied Behavioral Science, 35:
ioral Science, 22: 95–112. 479 –502.
Gricar, B., & Brown, L. D. 1981. Conflict, power, and organi- Lewis, J. D., & Weigert, A. 1985. Trust as a social reality.
zation in a changing community. Human Relations, 34: Social Forces, 43: 967–985.
877– 893.
Lovelace, K., Shapiro, D. L., & Weingart, L. R. 2001. Maximiz-
Gulati, R. 1999. Network location and learning: The influence ing cross-functional new product teams’ innovativeness
of network resources and firm capabilities on alliance and constraint adherence: A conflict communications
formation. Strategic Management Journal, 20: 397– 420. perspective. Academy of Management Journal, 44: 779 –
Gulati, R., Nohria, N., & Akbar, Z. 2000. Guest editors’ intro- 793.
duction to the special issue: Strategic networks. Strate- Maguire, S., Phillips, N., & Hardy, C. 2001. When “silence ⫽
gic Management Journal, 21: 203–215. death,” keep talking: Trust, control and the discursive
76 Academy of Management Review January

construction of identity in the Canadian HIV/AIDS treat- Salancik (Eds.), New directions in organizational behav-
ment domain. Organization Studies, 22: 287–312. ior: 1–54. Chicago: St. Clair.
McCann, J. E. 1983. Design guidelines for social problem- Schein, E. H. 1986. Organizational culture and leadership: A
solving interventions. Journal of Applied Behavioral Sci- dynamic view. San Francisco: Jossey-Bass.
ence, 19: 177–189.
Schweiger, D. M., Sandberg, W. R., & Ragan, J. W. 1986. Group
Milne, G. R., Iyer, E. S., & Gooding-Williams, S. 1996. Envi- approaches for improving strategic decision making: A
ronmental organization alliance relationships within comparative analysis of dialectical inquiry, devil’s ad-
and across nonprofit, business, and government sectors. vocacy, and consensus. Academy of Management Jour-
Journal of Public Policy and Marketing, 15: 203–215. nal, 29: 51–71.
Mohrman, S., Cohen, S., & Mohrman, A. 1995. Designing team Simonin, B. 1997. The importance of collaborative know-how:
based organizations. San-Francisco: Jossey-Bass. An empirical test of the learning organization. Academy
Mumby, D., & Clair, R. P. 1997. Organizational discourse. In of Management Journal, 40: 1150 –1174.
T. A. van Dijk (Ed.), Discourse as social interaction: 181– Swidler, A. 1986. Culture in action: Symbols and strategies.
205. London: Sage. American Sociological Review, 51: 273–286.
Palmer, I., & Dunford, R. 1996. Conflicting uses of metaphors: Tajfel, H. 1972. La categorisation sociale [Social categoriza-
Reconceptualizing their use in the field of organiza-
tion]. In S. Moscovici (Ed.), Introduction de la psychologie
tional change. Academy of Management Review, 21:
sociale, vol. 1: 272–302. Paris: Larousse.
691–717.
Taylor, J. R., Cooren, F., Giroux, N., & Robichaud, D. 1996. The
Parker, I. 1992. Discourse dynamics: Critical analysis for so-
communicational basis of organization: Between the
cial and individual psychology. London: Routledge.
conversation and the text. Communication Theory, 6:
Phillips, N., & Hardy, C. 1997. Managing multiple identity: 1–39.
Discourse, legitimacy and resources in the UK refugee
Taylor, J. R., & Van Every, E. J. 1993. The vulnerable fortress:
system. Organization, 4: 159 –185.
Bureaucratic organization in the information age. To-
Phillips, N., & Hardy, C. 2002. Discourse analysis: Investigat- ronto: University of Toronto.
ing processes of social construction. Thousand Oaks, CA:
Sage. Trist, E. 1983. Referent organizations and the development of
interorganizational domains. Human Relations, 36: 269 –
Phillips, N., Lawrence, T. B., & Hardy, C. 2000. Interorganiza- 284.
tional collaboration and the dynamics of institutional
fields. Journal of Management Studies, 37: 23– 43. Tsoukas, H., & Chia, R. 2002. On organizational becoming:
Rethinking organizational change. Organization Sci-
Phillips, N., Lawrence, T. B., & Hardy, C. 2004. Discourse and ence, 13: 567–582.
institutions. Academy of Management Review, 29: 635–
652. van Dijk, T. A. 1997. Discourse as interaction in society. In
T. A. van Dijk (Ed.), Discourse as structure and process:
Potter, J., & Wetherell, M. 1987. Discourse and social psychol-
1–37. London: Sage.
ogy: Beyond attitudes and behavior. London: Sage.
Warren, R., Rose, S., & Bergunder, A. 1974. The structure of
Powell, W. W. 1990. Neither market nor hierarchy: Network
urban reform. Lexington, MA: Heath.
forms of organization. Research in Organizational Be-
havior, 12: 295–336. Weick, K. E. 1998. Improvisation as a mindset for organiza-
tional analysis. Organization Science, 9: 543–555.
Powell, W. W., Koput, K. W., & Smith-Doerr, L. 1996. Interor-
ganizational collaboration and the locus of innovation: Weick, K. E., & Roberts, K. H. 1993. Collective mind in orga-
Networks of learning in biotechnology. Administrative nizations: Heedful interrelating on flight decks. Admin-
Science Quarterly, 41: 116 –146. istrative Science Quarterly, 38: 357–381.
Pratt, M. G., & Foreman, P. O. 2000. Clarifying managerial Wells, J. C., & Liebman, W. B. 1996. New models of negotia-
responses to organizational identities. Academy of Man- tion, dispute resolution, and joint problem solving. Ne-
agement Review, 25: 18 – 42. gotiation Journal, 12: 119 –136.
Psathas, G. 1995. Conversation analysis: The study of talk in Westley, F. 1990. Middle managers and strategy: The micro-
interaction. London: Sage. dynamics of inclusion. Strategic Management Journal,
Sabel, C. F. 1993. Studies of trust: Building new forms of 11: 337–351.
cooperation in a volatile economy. Human Relations, 46: Westley, F., & Vredenburg, H. 1991. Strategic bridging: The
1133–1170. collaboration between environmentalists and business
Saffold, G. S., III. 1988. Culture traits, strength, and organi- in the marketing of green products. Journal of Applied
zational performance. Academy of Management Re- Behavioral Science, 27: 65–90.
view, 13: 546 –558. Woodilla, J. 1998. Workplace conversations: The text of orga-
Salancik, G. R. 1977. Commitment and the control of organi- nizing. In D. Grant, T. Keenoy, & C. Oswick (Eds.), Dis-
zational behavior and belief. In B. M. Staw & G. R. course and organization: 31–50. Sage: London.
2005 Hardy, Lawrence, and Grant 77

Cynthia Hardy is professor of management in the Department of Management, Uni-

versity of Melbourne. She received her Ph.D. in organizational theory from Warwick
University, UK. Her research interests include power, organizational discourse, and
collaborative strategy. She is cofounder of the International Centre for Research on
Organizational Discourse, Strategy and Change.

Thomas B. Lawrence is the Weyerhaeuser Professor of Change Management at Simon

Fraser University in Vancouver, Canada. He received his Ph.D. in organizational
analysis from the University of Alberta. His research focuses on the dynamics of
power, change, and institutions in organizations and organizational fields.

David Grant is chair of organizational studies, The School of Business, University of

Sydney. He received his Ph.D. from the London School of Economics and Political
Science. His primary area of research interest is organizational discourse, especially
where it relates to organizational change initiatives.