“Sticking One's Neck Out”: Reducing Mistrust in Sri Lanka's Peace Negotiations

Why are some confidence‐building strategies more successful than others? Lack of trust has been widely used to explain the failure of peace negotiations (e.g., Gross Stein 1996; Kelman 1997; Walter 1999, 2002). In situations characterized by high levels of vulnerability, such as in internal armed conflicts, the warring parties have strong reasons to be distrustful of the other side's intentions. In addition, the antagonists often find it difficult to correctly interpret the other side's actions. Previous research on the resolution of internal armed conflicts has therefore highlighted the importance of third parties in reassuring the parties that negotiation is a viable option. However, we know less about the role of the belligerents themselves in bringing intrastate conflicts to an end.

This study illustrates the potential of the belligerents' actions to initiate peaceful interactions. For theory‐building purposes, we explore how different sorts of political risks can be taken in the communicative process to transform foes into partners in peacemaking. We do so by analyzing measures taken to reduce mistrust during two attempts to negotiate peace in Sri Lanka: one initiative undertaken in 1994 and another peace process initiated in 2002.¹

A theoretical framework of communicative signaling is applied to these two peace initiatives in order to develop propositions about why some mistrust‐reducing strategies succeed while others fail. As such, this article contributes to the understanding of strategies that can be used to build enough confidence between belligerents to elicit cooperative behavior and enable serious negotiations to take place.

Combining theories from social psychology with rational choice approaches, we argue that conflicting parties can strategically use risks, vulnerabilities, and costs inherent in a de‐escalation process to communicate conciliatory intentions. This means that obstacles for peace can be transformed into important catalysts for the reduction of mistrust. By taking measures that involve significant degrees of risk — including a potential loss of face, possible security risks, and the risk that critical information will be revealed — parties to a negotiation can credibly reveal nonoffensive intent.

To develop general propositions about the dynamics that can bring about a reduction in mistrust, we draw on a comparison of Sri Lanka's peace processes. The peace initiative in 1994–1995 never produced negotiations on any substantial issues, but in 2002 the parties were able to begin negotiations on a political solution to the conflict. The peace process of 1994–1995 ended with the resumption of civil war in April 1995 after only three months of cease‐fire. The negotiations focused largely on procedural questions and on the reconstruction of war‐torn areas, and the parties were unable or unwilling to make any substantial concessions, which could have created potential bargaining opportunities.

In contrast, in 2002 the parties altered their positions regarding the conflict's underlying issue, which is the status of the northern and eastern parts of Sri Lanka, and they reached several substantial agreements on ways to improve the humanitarian situation on the ground. Although the peace process has since then experienced numerous difficulties — such as cease‐fire violations, political impasse, and increased tensions between the parties — most observers would agree that the initial phase of the peace process was a remarkable success at least in terms of generating cooperative behavior and achieving substantial concessions from both sides.

A comparative analysis of the initial sequences of the two peace initiatives sheds some light on why the peace process reached a stalemate in 2003. The decision of one of the parties, the Liberation Tigers of Tamil Eelam (LTTE), to break away from the political negotiation process in the spring of 2003 significantly altered the process. Progress slowed, momentum was lost, and the process at the political level went into a stalemate. Initial optimism was transformed into rising levels of frustration, and verbal threats of a return to war were made. As we will suggest in this article, the shift corresponded with a departure from the strategy employed at the beginning of the peace process.

This article is divided into five parts. First, we discuss the importance of trust in negotiation processes and provide a theoretical framework that highlights confidence‐building strategies to handle the problem of mistrust in negotiations. We suggest how belligerents can deliberately impose costs on themselves as a way of demonstrating their interest in finding a peaceful solution to the conflict. Second, in a comparative analysis of the two latest peace initiatives in Sri Lanka, we apply this analytical framework and analyze how measures implemented during these processes were communicated by each side. Third, we discuss the theoretical implications of the analysis for how trust may be built in contexts characterized by high degrees of mistrust. Fourth, we outline the implications for peace in Sri Lanka and, in particular, discuss how the lack of conciliatory signaling may constitute an important obstacle for further progress toward a negotiated settlement. Finally, we summarize the main findings and discuss the wider theoretical implications for the study of trust in internal armed conflicts.

Note from the outset that our purpose in this article is not to account for all aspects of the dynamics of these particular negotiation processes but rather to understand how confidence‐building strategies can be improved and made more effective to help initiate meaningful negotiation after long and violent conflict. There are indeed additional significant variables that help explain the failure of the 1994–1995 peace process and the partial success of the 2002 process; these include the absence and presence of third‐party mediation, changes in the international context such as the Global War on Terror, and the degree of war weariness in the country.²

The concept of trust has received a great deal of attention in the literature on negotiation. In spite of this, few studies have studied trust in the context of conflict, in particular armed and violent conflict (Mitchell 2000). However, previous research on conflict resolution has indicated that mistrust is a significant obstacle to the resolution of armed conflict (Kelman 1997; King 1997; Kydd and Walter 2002; Walter 2002). Concepts such as Gradual Reciprocation in Tension Reduction, CBMs (Confidence‐Building Measures), and conciliatory signaling (see e.g., Kydd 2000a; Osgood 1962), have been used to study trust building in de‐escalation processes. Cases studied in depth include the Cold War (Kydd 2000b), the conflict over the Falkland Islands between Great Britain and Argentina (Mitchell 1991), and the Middle East conflict (Gross Stein 1991; Kriesberg 1992; Mitchell 2000).

Why is trust important in a negotiation process to end violent conflict? Trust is important because negotiating peace with the enemy is a risky endeavor. Trust “presupposes a situation of risk” (Luhmann 1998: 97). Most importantly, it is a risk in the sense that, as a group, the other party can exploit concessions made in the negotiation process. In such a situation, it is important that mistrust is reduced in order to begin meaningful negotiations. Or, as argued by Christopher Mitchell: “To begin a successful accommodative sequence, the high level of mistrust inherent in a relationship brought about by protracted and damaging conflict must be reduced, if only to a point where communications about possible non‐coercive solutions can be exchanged. . . . Greater trust is a necessary precondition to any progress toward any solution” (Mitchell 2000: 165).

After a long period of violence, however, it is not realistic to assume a high level of trust. Thus, for antagonistic relationships, confidence building is mainly about reducing the level of mistrust (Mitchell 2000). In addition, while cooperation does not require trust (Hoffman 2002; Kydd 2000a), the reduction of mistrust is important because it facilitates problem solving (Rubin, Pruitt, and Kim 1994; Schoppa 1999). Thus, a decline in mistrust between belligerents must occur before substantial negotiations about a prospective settlement can take place. The initiation and continuation of peace activities can thus be used as a proxy for determining if and by how much mistrust has been reduced.

The dilemma of mistrust applies to both international conflict as well as to conflicts within states. The problem of mistrust, however, is likely to be further aggravated in internal conflicts such as the one in Sri Lanka, where the parties must live side by side long after the fighting subsides (Posen 1993).

In this article, CBMs are studied as part of a larger set of communications between parties in a negotiation process. For a confidence‐building strategy to be successful, a party must demonstrate that it is trustworthy in its claimed willingness to pursue peace negotiations. But after a long and protracted conflict, the relationship between parties is more likely to be characterized by mistrust, fear, hatred, and other negative emotions and attitudes. Moreover, in a bargaining situation, the parties often have incentives to cheat and mislead for tactical reasons. For instance, conflicting parties may promise peaceful behavior in order to elicit rewards from third parties or in order to fool the other side into an exploitable position. Predictably, credible communication is thus difficult to achieve (Jönsson 1990; Schelling 1995).

A key question, then, is how the parties can credibly convey their willingness to negotiate peace, despite the presence of cognitive barriers that hinder communication as well as strategic incentives to misrepresent intentions? In line with previous thinking on the topic — both within social psychology and rational choice approaches — we explore the conditions under which the parties themselves can deal with the lack of trust through unilateral actions.³

As suggested by several scholars, parties can increase their credibility by taking measures that impose a substantial cost on themselves: a strategy that has been labeled the “sticking one's neck out” strategy (Mitchell 2000: 177). “A party seeking trustworthiness needs to place itself in a position of increasing its own losses as a result of its failure to fulfill verbal promises or commitments by subsequent deeds” (Mitchell 2000: 175). Or, formulated differently: “the trustworthy actor must do something that the untrustworthy actor would hesitate to do” (Kydd 2000a: 399). The cost will distinguish between those parties who are genuinely concerned about negotiating peace with the enemy and those who are negotiating with the other party for tactical reasons.⁴ In rational choice literature, this kind of communication is referred to as costly signaling — as distinguished from “cheap talk”— which may be used to persuade the other actors of their credibility (see e.g., Fearon 1992; Kydd 2000a). Thus, by taking a conciliatory initiative that is costly, conciliatory intentions can be credibly communicated.

What kind of costs can the parties take on in order to reveal conciliatory intentions in the context of armed conflict? We contend that the parties may strategically use potential costs, vulnerabilities, and risks inherent in peace processes as a way of increasing the credibility of the conciliatory communication. Jeffrey Rubin and his colleagues argued that there are three different kinds of potential costs involved in a de‐escalation process (Rubin, Pruitt, and Kim 1994). First, there are those costs where the parties risk losing image and prestige (image loss). Particularly important are measures where the parties may be portrayed as being weak. Second, there are costs related to losing the bargaining position, which may effect the final agreement (position loss). And third, there are costs related to loss of information regarding strength, commitment and other kinds of crucial information in the context of conflict (information loss). Building on Jeff Rubin, Dean Pruitt, and Sung Hee Kim's categorization, we will elaborate on three different kinds of costs in the Sri Lankan peace processes: image costs, security costs, and information costs.

In Sri Lanka, a long‐standing and violent conflict has been fought between the Tamil rebel group LTTE and the Singhalese‐dominated government.⁵ The conflict has its origin in the discriminatory measures imposed by the Singhalese majority against the Tamil minority in the post‐independence period. As the conflict became increasingly militant, demands for an independent Tamil state comprising part of the north and east of the island gained strength, and a civil war broke out in 1983. Since then, several attempts have been made to solve the conflict through negotiations.⁶

This section compares the confidence‐building strategies used by the LTTE and the government in 1994–1995 and 2002, respectively. Direct negotiations between the LTTE and the government were initiated in October 1994, following the election that brought Chandrika Kumaratunga and the People's Alliance (PA) to power. The negotiations, which took place in Jaffna, a Tamil‐dominated city in the northern part of Sri Lanka, resulted in a three‐month cease‐fire. But the parties had fundamentally different views about the negotiation process and failed to seriously negotiate a political solution to the conflict. Instead, the talks centered on how to improve the situation in the war‐torn areas and on the conditions for reaching a permanent cease‐fire. In April 1995, the LTTE abandoned the cease‐fire, and the period following the breakdown of the negotiations came to be the most violent period so far in the history of independent Sri Lanka.

A new initiative for negotiations was launched in 2002, after Ranil Wickremasinghe and the United National Party (UNP) came to power. A cease‐fire agreement (CFA) between the LTTE and the Sri Lankan government was reached in February 2002. Peace negotiations were then initiated in mid‐September 2002 — the first ones in seven years.

This time, substantial progress was made. During the negotiations, the parties considered potential solutions for the conflict focused on some kind of self‐determination for the Tamil‐dominated areas. At a press conference held in Oslo, Norway, at the beginning of December, the parties announced after only three rounds of talks that they had decided to “explore a solution founded on the principle of internal self‐determination in areas of historical habitation of the Tamil‐speaking people based on a federal structure within a united Sri Lanka” (BBC News, December 5, 2002). This statement should be seen in the light of the belligerents' previously held positions: the LTTE had demanded a separate state in the northeast of Sri Lanka, while the government had considered federalism anathema since the 1950s (de Silva 1998). Thus, the parties had agreed on a formula for the resolution of the conflict, which by itself was a historical event. The statement marked significant concessions from both sides, opening up possibilities for reaching a negotiated settlement.

Image costs arise from the risk that certain concessions will make a party appear weak — to both the opposing side and to members of its own constituency. Image and status are critical in a negotiation process. A loss of image (or “face”) may engender accusations of betrayal from within one's own party. A comparison between the two peace initiatives in Sri Lanka reveals some interesting contrasts regarding internal criticisms, the recognition of the other party as a legitimate actor, and media coverage.

In 2002, the Wickremasinghe government, to a larger extent than the Kumaratunga government in 1994, exposed itself to possible image loss when it initiated talks with the LTTE.⁷ Wickremasinghe faced more severe internal criticism than Kumaratunga, for three reasons.

First, in 1994 the governing party, the PA, held both the presidency and the post of prime minister,⁸ which made the party stronger and less likely to be challenged when it took conciliatory measures. By contrast, in 2002, Kumaratunga's party had lost the prime minister post (and subsequently the government) to Wickremasinghe, but Kumaratunga still remained president. Hence, the political leadership was more vulnerable in the latter peace process because any conciliatory measures could therefore be politically defied.

Second, in 1994, there appeared to be some degree of consensus between the UNP and the PA, which are the two main Singhalese parties, for the first time in a long time. Both parties seemed to recognize that for the conflict to end, the LTTE must be part of the solution (Shankar 1994). In 2002, the political leadership at the center was divided (Uyangoda 2005). President Kumaratunga was an outspoken critic of the design of the 2002 peace process, although she expressed her basic support for the process itself.

The third reason that peace talks posed a greater risk of image loss for government leaders in 2002 than they did in 1994 was that, by 2002, a weak anticonciliation opposition had grown stronger. Following the 1995 breakdown of talks, Janatha Vimukti Permuna (JVP), an extreme‐nationalist Singhalese party, grew in power. In 2002, a highly visible opposition challenged the peace process and accused the prime minister of dividing the country. On several occasions, the JVP organized demonstrations with militant overtures that involved thousands of Buddhist monks protesting the cease‐fire and the peace talks.⁹ Although nationalistic groups opposed to the peace process had existed in 1994, they lacked the mandate the JVP had by 2002, with sixteen seats in a 225‐member national parliament. Thus, by 2002, the critics of the peace initiative were more outspoken and powerful than they had been in 1994.

Clearly, the peace process faced more severe internal criticism on the government side in 2002 than it did in the mid‐1990s. Given the strength of resistance to a peace settlement, the fact that the government nevertheless initiated a negotiation process with the LTTE, can be seen as an indication of the government's commitment to peace negotiations.

The parties to a conflict can also expose themselves to image loss by recognizing the other party. Granting recognition is costly because rebels fight for recognition and legitimacy (Guelke 2003; Zartman 1995). Recognition, once offered, cannot easily be taken back. Moreover, recognition of the other party as a legitimate actor can also jeopardize a leader's standing within his/her own party. By cooperating with the enemy, a leader risks being perceived as traitor to the cause. The government's willingness to recognize the LTTE as an equal partner in the negotiation process is illustrated by the ways in which the negotiation teams were put together and in which reconstruction has taken place.

Clearly, Wickremasinghe's negotiation strategy in 2002 was to handle the LTTE as an equal and legitimate negotiating partner (Perera 2002). This effort was reflected in the composition of the negotiation teams. Wickremasinghe's government delegation was made up of high‐level politicians, drawn from the inner circle of the government, with substantial power and close relations to the prime minister. And the leadership of the Sri Lankan government's Secretariat for Coordinating the Peace Process (SCOPP) came from the foreign services, with the competence to handle negotiations with equal parties.

In contrast, in 1994, the PA government's unwillingness to recognize the LTTE as an equal partner in the negotiations was reflected in the way in which the negotiation team was put together. The members were low‐ranking and only two were government officials, and they lacked the experience to conduct serious negotiations at a governmental level. In addition, the delegates failed to make use of the experience and expertise of government negotiators from previous peace attempts (Schaffer 1999). Whatever the intentions behind this team selection may have been, its composition increased the perception on the LTTE side that the government's effort to resolve the conflict was not serious (Balasingham 2000: 145; Puleethevan 2002). Consequently, the design of the negotiation process gave little recognition to the rebel group. Another critical difference involved the reconstruction efforts. In 1994–1995 they were conducted as governmental projects, but in 2002 they were planned as joint ventures between the government and the LTTE. In 2002, the government's decision to give the LTTE a central role in reconstruction reinforced the perception of parity between the parties.

The restoration of normalcy in the war‐affected areas has been a long‐standing demand of the LTTE. In 2002, alleviating the harsh living conditions of the people in the North and East was given high priority by both the government and the rebels. Measures included opening the A9 highway to Jaffna, removing the embargo on products going to war‐affected areas, and providing major funding to a common rehabilitation fund. After an international donor conference in Oslo in December 2002, the LTTE and the government agreed to set up a joint commission to manage international funding.¹⁰ What is striking about all these developmental activities and reconstruction efforts is that Wickremasinghe achieved them all in cooperation with the LTTE. In this way the government did, in fact, strengthen the LTTE, because it would be the LTTE that was to deliver an increased living standard to the Tamil people.

In comparison, during the 1994 peace process, although the Kumaratunga regime negotiated directly with the LTTE, government officials publicly stated that the “peace talks would help to separate the Tamil people from the LTTE” (Perera 1998: 244). At the time, the LTTE had control over the North. With the initiation of the peace process, the LTTE feared marginalization, a fear that was reinforced by the public support among the Tamil people for the government, as had been seen in the election in 1994 that brought Kumaratunga to power. Spontaneous demonstrations in support of Kumaratunga also broke out in Tamil‐dominated Jaffna when the government negotiations teams arrived for talks. Kumaratunga grew popular in the North following her promise to end the conflict, and shops began selling “Chandrika bangles” and “Chandrika soap” (Sri Lanka Monitor 1994).

In 2002, the government risked losing face by recognizing the LTTE as an equal partner, which ultimately proved to be a successful mistrust‐reducing strategy. By meeting the LTTE on equal terms and by engaging in conciliatory, peace‐building, and humanitarian efforts in joint cooperation with the LTTE, the Wickremasinghe government recognized the rebels in a way the Kumaratunga government never did. Quite contrary, her efforts to build confidence were directed more toward the Tamil people and not toward the LTTE. Consequently, the rebel organization, with the ambition of being the “sole representative” of the Tamil people, feared being marginalized as a result of the peace process.

Extensive international media attention can also create a risk of image loss. Since 2002, both parties, but particularly the LTTE — as well as the conflict itself — have received more media attention than they did in 1994–1995, which stems partly from the increasing internationalization of the conflict.

The LTTE has made serious attempts to improve its tarnished international image. For example, in April 2002, the LTTE leader Velupillai Prabhakaran, who seldom gives interviews to the media, stepped forward, and — in civilian clothes rather than a military uniform — met the gathered international press in an LTTE‐held area in the north of Sri Lanka. More than three hundred local and international journalists attended the press conference. The message was clear: the LTTE welcomed a negotiated settlement to the conflict and, if a settlement meeting their aspirations was reached, would transform themselves into a political party. Prabhakaran's press conference conveyed to an international audience that the LTTE was seriously committed to peace and the restoration of normalcy. Given this effort to internationalize the peace process, backing down from its commitment to peace would seriously have hurt the image of the LTTE and thus made backing out of the process much riskier.

The parties were not similarly exposed to the glare of international attention during the 1994–1995 talks (Wijemanne 2002) and thus did not risk this particular loss of image. The exposure to the international community's attention, in general, and the international press and media, in particular, put the LTTE in a more vulnerable position in 2002 than in 1994–1995.

Thus, the image costs that the parties were willing to incur in terms of internal criticism, recognition of the other side, and media exposure were substantially higher in the 2002 process than they were in 1994–1995. With this in mind, we have developed the following proposition regarding mistrust‐reducing strategies:

When the parties reduce or give up some of their means to security they risk incurring security costs. Such risks are incurred when parties give up power and positions on the ground and when they open up access to territory. The Sri Lankan government took two measures during the 2002 peace process that stand in stark contrast to the 1994–1995 peace process: They lifted security barriers in Colombo, the capital, and elsewhere, and relaxed the embargo on LTTE‐held areas in the north and east of the country.

First, in a rather dramatic decision by the Wickremasinghe government, the security barriers in Colombo were lifted shortly after the signing of the cease‐fire. This measure was especially risky because of the potential it created for the LTTE suicide‐bombers, the so‐called “Black Tigers,” to come into the city. The CFA in 2002 also provided for the free movement of LTTE cadres dressed in civilian clothes in government‐held areas. Moreover, during the peace process, the LTTE was allowed to open political offices throughout the country. All these measures communicated good intentions and willingness to take risks for the sake of peace. The 1994–1995 process was quite different.

Relaxing the governmental embargo against the north and east had security implications. Increased resources, whatever their kind, became available that could have been used for military purposes and to strengthen the capacity of the LTTE. In 1994, the government had, indeed, relaxed the embargo, but the LTTE failed to interpret it as the gesture of goodwill that the government had intended for two reasons. First, the embargo was very limited, leaving a significant number of items on the list of prohibited products, which resulted in a scarcity of, for example, cement, an item necessary for rebuilding houses in the war‐affected areas (Uyangoda 1995). Furthermore, relaxation of the embargo was badly implemented on the ground, partly because of lack of cooperation from the army (Perera 1998).

The LTTE — seeing the discrepancies between the verbal declarations and the actions on the ground — concluded that the promise to lift the embargo during the 1994–1995 talks was insincere. But in 2002, the Wickremasinghe government's relaxation of the embargo was much more comprehensive and more fully implemented (Goodhand and Klem 2005). Thus, the government's willingness to take security risks for the sake of peace was not only declared, but confirmed.

The risks of incurring high security costs that both sides took during the 2002 peace process stand in stark contrast to what occurred during the 1994–1995 peace process. In 2002, parties, especially the government side, voluntarily took security risks and increased their vulnerability and chances of being exploited by the opponent. This leads us to the formulation of a second proposition:

Parties risk incurring information costs when they reveal information — about strength, commitment, and resources — that, by being revealed, could give the opposing party greater bargaining leverage. In the 2002 peace process in Sri Lanka, the parties made two significant changes related to monitoring by a third party and the organizational design of the negotiations, measures that increased transparency and revealed important information. Measures on the same scale were not taken during the 1994–1995 peace process.

In 2002, the parties invited representatives from the Nordic countries to supervise the cease‐fire. The parties had originally asked Norway to monitor the cease‐fire, but Norway was reluctant to take on the sole responsibility for this task. Hence, an agreement was reached to also include Sweden, Finland, Denmark, and Iceland in the monitoring team. Known as the Sri Lanka Monitoring Mission (SLMM), these representatives ensured that the process would be transparent by monitoring, reporting, and disseminating information about the activities of the parties on the ground. SLMM's assignment has been to observe any violation of the cease‐fire and to facilitate the resolution of disputes over implementation.¹¹ The mission consists of personnel drawn from the Nordic countries who are located in six districts of Sri Lanka, with headquarters in Colombo. Local monitoring committees comprise these European monitors as well as representatives from the government and the LTTE. Their task is to “conduct international verification through on‐site monitoring of the fulfillment entered into.”¹²

The Cessation of Hostilities agreement of January 1995 also made provisions for a monitoring team that was to include representatives from “Canada, Norway, Netherlands, ICRC [International Committee of the Red Cross] and from among retired judges or public officers, religious heads and other leading citizens” (Cessation of Hostilities Agreement as cited in Rajanayagam 1998: 204). But, unlike the 2002 monitoring mission, the earlier mission lacked a clear command structure. Following the agreement, foreign monitors arrived to oversee its implementation, but never resumed its supervision because the parties came into conflict over the work of the monitoring committees.¹³ Therefore, the mission was unable to build confidence between the parties.

Although the actual negotiations have been conducted in secrecy, the organizational design of the peace talks in 2002–2003 enhanced transparency between the parties. In particular, the government took deliberate measures to make the coordination of decision‐making related to the peace process more effective by establishing SCOPP. As a result, decision‐making on the part of the government became more transparent. Similarly, the LTTE created its own Peace Secretariat for coordination of the peace process activities, with headquarters at the political office in Kilinochi.

The Sri Lankan government's decision to create SCOPP was a response to the failures of previous peace initiatives, including the 1994–1995 process. At that time, peace process activities lacked a coherent central organization, with political and administrative decision‐making divided between various departments (coordinated by the president's office) and was therefore harder to oversee (Gooneratne 2002).

To conclude, two measures that distinguished the latter peace process from the former ones included the acceptance of third‐party monitoring and the creation of new organizational structures that increased transparency. Both third‐party monitoring and centralization of the administrative structure for peace negotiations exposed the parties to significant risks of incurring information costs. This leads us to our third proposition regarding mistrust‐reducing strategies:

To send olive branches after years of death and destruction is a process naturally marked by immense difficulties. Adversaries confront serious barriers and severe obstacles as they seek to communicate a message of conciliation to the other side. The turn from military conflict to a process of de‐escalation necessarily challenges each side's long‐held perceptions and emotional attitudes concerning its enemy (Gross Stein 1996; Kelman 1997). In particular, because of attributional distortion, conflicting parties are inclined to see cooperative behavior as a response to situational factors (Kelman 1997) rather than as a choice freely made by the enemy. In addition to cognitive barriers, the belligerents have strategic incentives to misrepresent information in the communication process. Periods of de‐escalation can be used to regroup and rearm, to manipulate support from the outside world, and to avoid revealing offensive plans and ambitions. The parties are aware of the other side's interest in exploiting the opportunities that the de‐escalation process creates. Because of these fears, they will be reluctant to send conciliatory signals and inclined to neglect indications of conciliatory intent by the other side.

Efforts at conciliatory signaling to reduce mistrust are components of de‐escalatory strategies that face major obstacles in internal armed conflicts. What inferences can be made from our exploration of Sri Lanka's negotiations? Our basic argument in this article is that mistrust can be reduced when parties reveal their conciliatory intent by exposing themselves to costs and risks that they would not be ready to accept if their intentions were less conciliatory. What does such “conciliatory risk‐taking” mean in the context of intrastate, armed conflict?

In this article, three basic types of costs or risks involved in a de‐escalation process (Rubin, Pruitt, and Kim 1994) were identified as obstacles for progress toward peace. However, we have argued that the belligerents can also use these costs strategically in order to communicate their conciliatory intentions. The negotiations in the Sri Lanka case illustrate how the risks involved in peacemaking can be managed creatively as part of the communications between the parties. Parties interested in reducing the enemy's skepticism and distrust may take measures that involve the risk of losing face (or image), that incur security risks, or that reveal potentially valuable information about the strength and commitment of the parties. Hence, the obstacles for peace may be transformed into important catalysts for the reduction of mistrust.

As we have argued above, the two peace processes were different in that the parties showed more readiness to risk their image and reputation in the 2002 process than in 1994–1995. For instance, by initiating and responding to conciliatory moves, the parties exposed themselves to internal criticisms and took the risk of being perceived as weak in front of the enemy. Leaders who are involved in de‐escalation may, in such circumstances, be accused of betraying their own cause. The strength of those who oppose compromise determines the level of the internal image costs.

It is interesting to relate the issue of dissent to the debate on “spoilers.” Spoilers frequently arise in response to a nascent peace process and are unwilling to accept a peace settlement that runs counter to their interests (Stedman 1997). Because of their interest in disrupting a peace process — most extremely by using violence — spoiler management has been singled out as one of the key activities for international and local actors (Licklider 2001; Stedman, Rothchild, and Cousens 2002). According to the logic applied in this article, however, the presence of potential spoilers in a negotiation process does not necessarily have a one‐sided negative effect on peacemaking. Instead, the willingness by the parties to initiate negotiations in spite of spoilers can convey a powerful conciliatory message (see also Hoddie and Hartzell 2005).

The two peace processes examined here were also different in terms of the security risks that were incurred. In 2002, the government was prepared to take security risks in order to reveal its conciliatory intentions. Taking such risks increases the credibility of one's commitment because it gives the opponent an opportunity to exploit an increased vulnerability in order to gain strategic advantages. However, in case there is a fallback to armed clashes, the downscaling of military measures may turn out to have been extremely costly. For instance, when hostilities resumed in Sri Lanka in 1990, following a period of negotiations between Sri Lankan President Ranasinghe Premadasa and the LTTE, hundreds of unprepared government soldiers died at the hands of the LTTE. To build confidence, Premadasa had made several concessions to the LTTE in the military realm, such as dismantling military camps. The LTTE had used the lapse in fighting and the concessions made by Premadasa to prepare for renewed fighting (Ranatunga 1998).

The third category of risk that we have elaborated upon in this article is the risk of revealing valuable information. It is generally costly to increase transparency because it reveals to their opponents vital information about the party's strength and commitments. Opening up the territory for access, control, and monitoring by the enemy or a third party risks revealing information about military strength and where the forces are located. This kind of information is useful for the opponent if a war relapses. The presence of third parties may reveal information about the intentions and commitment of the parties. The decision to allow third parties to enter into the negotiation process can thus signal information about the resolve of the belligerents and the extent of conciliatory intentions. As argued by Barbara Walter, “outside intervention provides important information about what type of opponent each side is facing, distinguishing predators from non‐predators. Domestics groups who are intent on aggression are unlikely to accept outside interference since this would jeopardize their ability to carry out any malicious plans” (Walter and Snyder 1999: 306). While this statement relates to military intervention, it is reasonable to assume that similar arguments can be applicable for civilian observer missions such as the SLMM.

The parties seem to have strategically used all these risks in the 2002 peace process to signal conciliation. This seems to have resulted in a lower level of mistrust between the former belligerents. Hence, the LTTE and the government “stuck their necks out” by taking costly measures for the sake of peace. This strategy can be seen in stark contrast to the 1994–1995 peace process where no such measures were taken, or, if taken, did not involve as much risk for either of the parties. Measures taken aimed at building trust in the mid‐1990s were generally considered “cheap” and were therefore unable to reduce mistrust between the belligerents, to the extent that serious negotiations and substantial concession making could begin.

This article has compared the 1994–1995 peace process with the process up to the end of 2002. Can this analysis help us understand developments in the peace process in Sri Lanka since then? What are the implications for peacemaking in this increasingly troubled nation?

The confidence built in the first year of the peace process significantly diminished during the first months of 2003. The timing of the turning point of the peace process is interesting: It coincided with the Sri Lankan government's financial meeting with international donors in Washington, DC in April 2003. The LTTE protested this meeting vigorously because they were prohibited from attending due to the U.S. designation of the LTTE as a terrorist organization. The government's decision to nonetheless go to Washington, DC raised serious questions about whether the government recognized the LTTE as an equal partner in the process. It was when this equality between the parties was threatened that the political process reached a deadlock.¹⁴ Granting the rebels a sense of parity was, as we have argued, a costly sign of conciliatory intentions from the government. When this equality was threatened, the emerging trust between the former antagonists started to evaporate, and consequently the peace process ran the risk of breaking down.

In the initial stages of the 2002 peace process, the LTTE exposed itself to international media, initiated organizational changes, and opened up some of its territory to the outside world, but its behavior in the later stages of the peace process has raised serious questions about its interest in transforming to a peaceful and democratic party. Although the guerilla group initially committed itself to seeking a negotiated settlement, its continued reliance on violence as a means to dispose of political enemies and opponents has made that commitment less credible. Politically motivated assassinations and forced recruitment of underage Tamils as well as other violations of human rights have generated legitimate concerns — in Sri Lanka and internationally — about the LTTE's interest in peace and democratic development. These concerns have resulted in a number of states and organizations banning the LTTE as a terrorist entity, including Canada and the European Union. Hence, the conciliatory image of the LTTE that emerged as the organization took measures with substantial political costs in the early stages of the peace process began to fade away. It would be costly for a military organization such as the LTTE to accept dissent and to rely purely on consent when recruiting their cadres. By using the logic explored in this article, we suggest that by exposing themselves to these costs, the LTTE can find ways to credibly reveal peaceful intentions in their communication with the government side.

The analysis above also gives some insights into possible strategies that the Sri Lankan government could use to reduce the LTTE's skepticism of reaching a deal. By challenging the opponents of peaceful coexistence, most prominently the JVP and the Jathika Hela Urumaya (JHU),¹⁵ the government could signal a commitment to peace. Against the backdrop of increasingly outspoken and influential opposition to negotiated settlement with the LTTE, it would be politically costly to nevertheless initiate negotiations and problem‐solving endeavors with the rebels.

A recent example concerns the government's willingness to accept a joint mechanism to distribute aid to reconstruction following the tsunami that hit Sri Lanka in December 2004. The joint mechanism awards the LTTE with legitimacy by giving them a role in decision‐making over the tsunami recovery. The proposal was vigorously contested by the JVP and led to the party's resignation from the governing coalition. As put by one analyst: “the government's willingness to sign the P‐TOMS (Post‐Tsunami Operational Management Structure) agreement with the LTTE in the face of formidable opposition is taken as proof of the government's commitment to the peace process” (Perera 2005).

Strategies to reduce mistrust are sorely needed in light of the deteriorating security situation in Sri Lanka. More direct attacks between the rebels and government forces, alarming acts of violence perpetrated against innocent civilians, and an increasing rivalry between Tamil groups have escalated the conflict in 2006. In fact, as this goes to press (June 2006), most analysts expect that there will be a return to full‐fledged civil war in Sri Lanka.

Moreover, politically, the peace process has been stalemated since 2003, and few expect any advancement toward mutual accommodation in the near future. At this point, it is critical that the parties find ways to reinvigorate a robust cease‐fire and initiate new negotiations. This will require the building of trust, which can be achieved though substantial risk taking for the sake of peace.

In the 2002 peace process, the parties were more successful in initiating substantial negotiations toward a political solution to the conflict in Sri Lanka than they were in 1994–1995. The conflicting parties showed readiness to counter internal criticism, to expose themselves to increased media attention, and to recognize the other party as an equal. In addition, they implemented measures that heightened security risks for themselves. The willingness to take such calculated and costly measures from both sides reduced mistrust between the parties and created favorable conditions for peace negotiations. De‐escalation processes leave parties vulnerable and may serve as obstacles for progress toward peace. However, as we have shown in this article, if the parties use these obstacles in innovative ways, the obstacles for peace may be transformed into important catalysts for the reduction of mistrust. This can lay the ground for further progress toward peace.

The analysis has wider theoretical implications. The first relates to how trust is created among former belligerents. Given the high level of mistrust that characterizes the relationship between parties in internal armed conflicts, previous research has stressed the importance of “guarantors” for easing the dilemmas created by mistrust. Guarantors are third parties who are willing and able to credibly enforce cooperative behavior between the former foes. They are needed to reassure that exploitative behavior will be punished, should one of the sides renege on mutual agreements (Walter 2002). Whereas previous research has been dominated by a focus on the role that third parties may play in bringing armed conflicts to an end, this study instead illustrates the potential of the actions by the belligerents themselves to bring about de‐escalation. Clearly, Norway has been important in bringing the parties together in the Sri Lankan peace process. However, the 2002 peace process was primarily driven forward by the actions and measures taken by the LTTE and the government, and by the dynamic between them.

The second implication of the analysis concerns the context in which conciliatory signaling can be a useful catalyst for negotiations to begin. Conciliatory signaling is not a novel idea in academic writing. The current theoretical debate has its origin in Charles Osgood's work during the days of the Cold War. Osgood suggested that the U.S. should take unilateral initiatives aimed at reducing the tension between the contesting superpowers and thereby reduce the risk of global warfare (Osgood 1962).

Whereas conciliatory signaling has been discussed mostly in relation to interstate conflicts, the Sri Lankan case shows the usefulness of this framework for understanding the dynamics of de‐escalation in internal armed conflicts. The application of conciliatory signaling to conflicts within states, however, is not without its critics. Barbara Walter dismisses the possibilities of using these tactics in the context of conflicts between governments and insurgent groups because intrastate security dilemmas are so immense and acute. Using costly, conciliatory signals, she argues, would “either expose the sender to such danger that even peace‐loving groups would avoid using them or they are too easy to mimic by more Machiavellian groups to have the desired effect of relaying peaceful intentions” (Walter 1999: 136). On the other hand, Matthew Hoddie and Caroline Hartzell (2005) show how costly, conciliatory signaling was used in the internal conflict over Mindanao during the Philippine peace process in the mid‐1990s. In line with their findings, we suggest that conciliatory signaling may indeed be an important mechanism for initiating meaningful peace negotiations to solve internal, violent conflict.