Professional Documents
Culture Documents
Melander
✣ Erik Melander
T
he escalation of the Nagorno-Karabakh conºict in 1988–1992
from a non-violent political struggle to a full-scale ethnic war was a momen-
tous development in the ªnal phase of the Cold War. Some analysts have sug-
gested that this local conºict gravely weakened the Soviet Union and thus di-
rectly contributed to the end of the Cold War. The Nagorno-Karabakh conºict
has also been depicted as part of a general wave of ethnic violence that arose
once the Cold War was largely over. Previous scholarly work has tended to por-
tray the escalation of the Nagorno-Karabakh conºict as the result of a surge of
nationalist sentiment unleashed under the more relaxed regime of Mikhail
Gorbachev. According to this interpretation, the path to war was largely un-
avoidable after deadly ethnic riots occurred in Sumgait in February 1988.1
A closer examination of new primary evidence, including interviews with
key participants, does not bear out these earlier views. Crucial developments
in the summer of 1991 indicate that there was a possibility to avert the war at
that time, more than three years after the events in Sumgait. In 1991 the po-
litical leadership of the Armenian separatist movement in Nagorno-Karabakh
offered the Azerbaijani authorities a virtual capitulation in exchange for a ces-
sation of hostilities. The war broke out because an alternative, more radical
Armenian leadership gained the upper hand in Nagorno-Karabakh during the
abortive August 1991 coup in Moscow, when the central authorities were
temporarily disengaged.
1. See for example Ronald Grigor Suny, “Nationalism and Democracy in Gorbachev’s Soviet Union:
The Case of Karabagh,” in Rachel Denber, ed., The Soviet Nationality Reader: The Disintegration in
Context (Boulder: Westview Press, 1992), p. 493; Suzanne Goldenberg, Pride of Small Nations: The
Caucasus and Post-Soviet Disorder (London: Zed Books, 1994), p. 162; Svante Cornell, “Undeclared
War: The Nagorno-Karabakh Conºict Reconsidered,” Journal of South Asian and Middle Eastern
Studies, Vol. 20, No. 4 (1997), p. 6; and Svante Cornell, Small Nations and Great Powers (Richmond:
Curzon Press, 1999).
Journal of Cold War Studies
Vol. 3, No. 2, Spring 2001, pp. 48–75
© 2001 by the President and Fellows of Harvard College and the Massachusetts Institute
of Technology
48
The Nagorno-Karabakh Conflict Revisited
49
Melander
have been resolved with much less bloodshed if the actors had been able to see
through the strategic uncertainties of the situation. Incomplete information
about capabilities and resolve are therefore an indispensable part of the expla-
nation for why this war began.
More speciªcally, if the successive regimes in Azerbaijan had been able to
foresee the ultimate dissolution of the Soviet Union and the catastrophic con-
sequences of Russia’s disengagement and its subsequent reorientation in favor
of Armenia, they would reasonably have preferred to cut their losses rather
than use force to resist Armenian separatism and irredentism. On the other
hand, if the political leadership of the Armenian separatist movement in
Nagorno-Karabakh had expected the kind of brutal response by the Soviet
Army that took place under “Operation Ring” in the spring of 1991, these
leaders would have had a strong incentive to retract their demands. Had they
done this, they would have spared themselves the pains of launching a failed
initiative in the summer of 1991 that cost them their leadership positions and
was the likely reason for the assassination of Valerii Grigoriyan by Armenian
hard-liners. If the abortive coup in Moscow in August 1991 had not diverted
the attention of the Soviet leadership at this critical juncture, the Armenian
initiative might ultimately have succeeded in averting a full-scale war. This ar-
ticle shows that Kaufman’s assessment of the escalation in Nagorno-Karabakh
as “mass-led” rather than “elite-led” must be qualiªed; the struggle for pri-
macy between competing elites played a decisive role.5
The ªrst section of this article provides background information on the
Nagorno-Karabakh conºict and the main actors involved. The second section
discusses previous theoretical analyses of the causes of the war and proposes a
complementary perspective that focuses on strategic uncertainty as the obsta-
cle to negotiated outcomes. The third and fourth sections describe how the
conºict escalated. The fourth section pays particular attention to the Arme-
nian capitulatory initiative in the summer of 1991. The account here is based
largely on new primary evidence in the form of interviews with participants,
as well as published sources in Russian that have been previously overlooked.
The conclusion lays out the theoretical ªndings of the article.
50
The Nagorno-Karabakh Conflict Revisited
Michael P. Croissant, The Armenia-Azerbaijan Conºict: Causes and Implications (Westport: Praeger
Publishers, 1998); and Kaufman, “Ethnic Fears.”
7. The percentages are found in A. N. Yamskov, “Ethnic Conºict in the Transcaucasus: The Case of
Nagorno-Karabakh,” Theory and Society, Vol. 20, No. 5 (October 1991), pp. 631–660, esp. p. 645.
The background information provided here is based on the following works: Cornell, Small Nations;
Croissant, Armenia-Azerbaijan Conºict; Goldenberg, Pride of Small Nations; Kaufman, “Ethnic Fears”;
Mark Malkasian, “Gha-Ra-Bagh!” The Emergence of the National Democratic Movement in Armenia
(Detroit: Wayne State University Press, 1996); V. G. Mityayev, “Karabakhskii konºikt v kontekste
mezhdunarodnykh otnoshenii,” in E. M. Kozhokina, ed., Armeniya: Problemy nezavisimogo razvitiya
(Moscow: Rossiiskii Institut Strategicheskikh Issledovanii, 1998), pp. 487–554; and Alexei Zverev,
“Ethnic Conºicts in the Caucasus 1988–1994,” in Bruno Coppieters, ed., Contested Borders in the
Caucasus (Brussels: VUBPRESS, 1996). As far as possible, controversial information has been checked
against several different sources. See Norman K. Denzin and Yvonna S. Lincoln, Handbook of Qualita-
tive Research (Thousand Oaks: Sage Publications, 1994), p. 373; and Robert K. Yin, Case Study Re-
search: Design and Methods, Vol. 5 of Applied Social Research Methods Series (Newbury Park: Sage
Publications, 1989).
8. Goldenberg, Pride of Small Nations, p. 161; Mityayev, “Karabakhskii konºikt”; Yamskov, “Ethnic
Conºict in the Transcaucasus”; Zverev, “Ethnic Conºicts in the Caucasus”; and Kaufman “Ethnic
Fears,” p. 15f.
51
Melander
52
The Nagorno-Karabakh Conflict Revisited
Theoretical Perspectives
Conºict theory and theories of ethnic violence provide the best tools for un-
derstanding the warfare in Nagorno-Karabakh. Most theories of violent
12. Malkasian, “Gha-Ra-Bagh!” p. 120f; and Zverev, “Ethnic Conºicts in the Caucasus.”
13. Cornell, Small Nations, p. 353f; Malkasian, “Gha-Ra-Bagh!” p. 116ff; Rutland, “Democracy and
Nationalism”; and Zverev, “Ethnic Conºicts in the Caucasus.”
14. Malkasian, “Gha-Ra-Bagh!” p. 191; Georgiy Shakhnazarov, Tsena svobody: Reformatsiya
Gorbacheva glazami ego pomoshchnika (Moscow: Zevs, 1993), ch. 12; and Mikhail Gorbachev, Mem-
oirs (New York: Doubleday, 1995), ch. 15.
15. Robert D. Putnam, “Diplomacy and Domestic Politics: The Logic of Two-Level Games,” Interna-
tional Organization, Vol. 42, No. 3 (Summer 1988), pp. 427–460.
53
Melander
conºict fall into three categories: those that look at the actors’ willingness to
resort to violence; those that focus on opportunities to pursue a strategy of vi-
olence; and those that look at a combination of willingness and opportunity.
In addition, a new approach to conºict theory argues that the willingness-
opportunity paradigm does not sufªciently explain costly wars. This new ap-
proach, sometimes referred to as “strategic choice,” emphasizes the impact of
uncertainty and incomplete information on decision making.16
Both Donald Horowitz and Walker Connor contend that willingness is
the most important factor in instigating ethnic conºict. Horowitz looks at the
effects of ethnic anxiety and ethnic entitlement on increasing tensions, and
Connor stresses the importance of the family bonds associated with ethnic
identity in arousing ethnic passions. In contrast, Charles Tilly focuses on the
role of shared and collectively controlled opportunities in the competition for
social niches.17 Most theorists of ethnic conºict, especially those afªliated
with the Majority at Risk project—perhaps the best-known body of system-
atic research on the causes of ethnic conºict—refer to both willingness and
opportunity.18
In a thorough study of the origins of the Nagorno-Karabakh war, Stuart
Kaufman looks at willingness and opportunity of a particular kind. He argues
that willingness to ªght is generated by ethnic prejudice that gives rise to fears
of extinction among an ethnic group. If members of the group perceive such a
threat, they will take the opportunity to mobilize militarily if they believe they
are in no danger of central state coercion. The result, according to Kaufman,
is an ethnic security dilemma. Both sides in the ethnic conºict are driven by
their perceived security needs to take measures that threaten their opponents.
Unless a strong state resorts to severe repression against one or both sides
(what Kaufman refers to as the Stalinist model), or undertakes a more modest
form of repression (what Kaufman refers to as the Northern Ireland model),
the ethnic security dilemma will result in ethnic war.19
54
The Nagorno-Karabakh Conflict Revisited
55
Melander
56
The Nagorno-Karabakh Conflict Revisited
adversary were willing and able to wage war over Nagorno-Karabakh, why
did the conºict gradually escalate for almost four years before full-scale war-
fare erupted?23 This latter question seems especially puzzling if we bear in
mind the perceived advantages of striking ªrst in the Nagorno-Karabakh
conºict.24
A deceptively simple answer to these questions is that the parties pos-
sessed incomplete information about crucial aspects of the situation. For ex-
ample, neither side knew whether its opponent would succeed in imposing its
will through force. Neither side knew whether the Soviet authorities would
come down on one side or the other, or whether the central government
would merely withdraw from the conºict. In strategic choice theory these and
similar uncertainties will determine whether the outcome of a conºict is a
peaceful, negotiated settlement or war. To the extent that the parties hold
conºicting beliefs about the balance of resolve and the balance of power, the
search for a less costly negotiated solution may be impeded. For example, if
both sides in a conºict believe they will prevail relatively easily on the bat-
tleªeld, perhaps because they both expect support from a third party, then
neither will be willing to make any substantial concessions. If one side under-
estimates the resolve of its opponent, it is likely to demand more than the op-
ponent will concede. But if the parties overestimate each other’s resolve or
strength, they may be willing to seek a compromise.25
This relationship between the perceived balance of resolve and power on
the one hand, and the perceived range of acceptable divisions of the disputed
stakes on the other, implies that the parties in conºict face incentives to exag-
gerate their resolve and power. Through displays of force and other signals,
each side can try to intimidate the other into conceding more of the stakes.
(Sometimes, however, the parties have incentives to downplay their resolve
and power, perhaps to attract sympathy and support from third parties or
to prepare the ground for a surprise attack.) Since each side knows that the
other is inclined to conceal its true resolve, neither believes what the other
says. This uncertainty is likely to affect each side’s decision about whether to
escalate.
23. The length of the escalatory process could be attributed to the level of repression used by the cen-
tral government. But a more important factor was the uncertainty about how Moscow would respond
to a conºict. As will be shown in the empirical section, the uncertain and sometimes contradictory ex-
pectations about the reaction of the central authorities to different actions affected the behavior of all
parties involved.
24. Barry R. Posen, “The Security Dilemma and Ethnic Conºict,” Survival, Vol. 35, No. 1 (Spring
1993), pp. 27–47.
25. When all parties in a conºict correctly assess the balance of resolve and power, they should be able
to predict the outcome of a violent confrontation. If so, they should also be able to agree on conces-
sions mutually preferable to a costly and violent conºict.
57
Melander
Developments up to 1991
The status of Nagorno-Karabakh was in dispute both before and during the
Soviet era. The most recent Nagorno-Karabakh conºict began shortly after
Gorbachev announced his reform programs of perestroika and glasnost. In
1987 Armenians repeatedly petitioned the authorities in Moscow to transfer
Nagorno-Karabakh to Armenia. Soviet leaders mostly ignored these appeals
until the Armenian petitioners and demonstrators grew bolder. The demon-
strations and strikes went beyond the level of behavior that was normally tol-
erated, but they did not necessarily guarantee that violence would occur. Until
the early 1990s most Armenians believed that a negotiated solution to the
Nagorno-Karabakh conºict was still possible.
In late 1987 and early 1988, however, a series of tragic events drastically
altered the situation. By the end of 1987 Azeri refugees began to ºee the
NKAO.26 On 22 February Azeri demonstrators marched on Stepanakert to
protest a resolution adopted by the Nagorno-Karabakh provincial council ap-
pealing a transfer of the NKAO to Armenia. Just inside the border of the
26. Audrey L. Altstadt, “O Patria Mia: National Conºict in Mountainous Karabagh,” in W. Raymond
Duncan and G. Paul Holman, eds., Ethnic Nationalism and Regional Conºict (Boulder: Westview
Press, 1994); Cornell, Small Nations; and Kaufman, “Ethnic Fears,” p. 16.
58
The Nagorno-Karabakh Conflict Revisited
NKAO, near Askeran, Azeri demonstrators clashed with police and local Ar-
menians, and two Azeris were killed. When these casualties were announced
on Azerbaijani radio, violence against Armenians erupted in the town of
Sumgait (near Baku) on 27 February. Some of the 2,000 Azeri refugees who
had ºed to Sumgait were reported to have played a signiªcant role in this eth-
nic riot, which killed 26 Armenians and 6 Azeris, according to ofªcial
sources.27
The cause of the Sumgait riot continues to be a puzzle.28 It is clear, how-
ever, that the riot increased fears within the Armenian community.29 Al-
though these fears had an important long-term effect on the development of
the conºict, most Armenians during the early phase of the struggle for the
uniªcation of Nagorno-Karabakh with Armenia still believed in the potential
of a nonviolent political campaign. They were convinced that Moscow, and
Gorbachev personally, would be won over to their side.
From early on, however, a few more militant Armenians began to prepare
for an armed struggle in response to the events in Sumgait.30 Among them
were Igor Muradyan, one of the original members of the Karabakh Commit-
tee in Armenia, who had been expelled from the Committee because of his
27. Armenian sources claim a much higher number of Armenian casualties. See Joseph R. Masih and
Robert O. Krikorian, Armenia: At the Crossroads (Amsterdam: Harwood Academic Publishers, 1999),
p. 8.
28. Several alternative explanations can be proposed for the Sumgait events. One is that the deadly
ethnic riot was a spontaneous or locally instigated outbreak of uncontrolled violence, which the re-
publican and central authorities failed to prevent because of incompetence and inefªciency rather
than malign intent. A variant of this explanation holds that local elements in the Azeri maªa wanted a
cover under which Armenian competition could be eliminated. See Richard Sakwa, Gorbachev and His
Reforms 1985–1990 (Hempel Hempstead: Philip Allan, 1990), p. 244. See also Kaufman, “Ethnic
Fears.” Another possible explanation is that the events were organized by the Azerbaijani republican
authorities, possibly even in conspiracy with the central authorities in Moscow, as a warning to the se-
cessionist Armenians in NKAO. Some claim that before the riots the authorities in Baku threatened
the Armenians in Nagorno-Karabakh with violence unless the demands for uniªcation with Armenia
were withdrawn. See Vagan Bakhshiyevich Arutyunyan, Sobytiya v Nagornom Karabakhe. Khronika.
Chast’ 1. Fevral’ 1988 g.–Yanvar’ 1989 g. (Erevan: Izdatel’stvo AN Armyanskoi SSR, 1990). Yet an-
other possibility is that anti-perestroika forces organized the deadly ethnic riot to discredit Gorbachev
and his reforms. This was the explanation offered by Gorbachev himself. See Croissant, Armenia-
Azerbaijan Conºict, p. 29. Others suggest that Moscow or the U.S. Central Intelligence Agency sought
to destabilize the region for their own purposes, or that the riot was instigated by Armenian provoca-
teurs to justify Armenian nationalist policies. See Levon Chorbajian, Patrick Donabedian, and Claude
Mutaªan, The Caucasian Knot: The History and Geopolitics of Nagorno-Karabagh (London: Zed Books,
1994), p. 189. Finally, some Armenians believe there is a genetic predisposition among ethnic Turks to
murder Armenians. This view was encountered in several conversations during ªeldwork; see also Ida
Babayan, Armenia and Mountainous Karabakh on the Political Scene of the World (Erevan: Aghvank,
1995).
29. Goldenberg, Pride of Small Nations, p. 154; Edmund M. Herzig, “Armenia and the Armenians,” in
Graham Smith, ed., The Nationalities Question in the Post-Soviet States (London: Longman Group,
1996), p. 257; Malkasian, “Gha-Ra-Bagh!”; Masih and Krikorian, Armenia: At the Crossroads;
Mityayev, “Karabakhskii konºikt,” p. 490; and Zverev, “Ethnic Conºicts in the Caucasus.”
30. Masih and Krikorian, Armenia: At the Crossroads, p. 20f.
59
Melander
60
The Nagorno-Karabakh Conflict Revisited
33. Ibid., pp. 103–129. See also Goldenberg, Pride of Small Nations, p. 142f; and Herzig, “Armenia
and the Armenians,” p. 258f.
34. Christopher J. Walker, Armenia and Karabagh: The Struggle for Unity (London: Minority Rights
Publications, 1991), p. 126.
35. See the statement by NKAO First Secretary Poghosyan, quoted in Malkasian, “Gha-Ra-Bagh!”
p. 119.
61
Melander
36. Rutland, “Democracy and Nationalism”; Herzig, “Armenia and the Armenians,” p. 256; Crois-
sant, Armenia-Azerbaijan Conºict, p. 25f; Gerard J. Libaridian, The Challenge of Statehood: Armenian
Political Thinking since Independence (Watertown, MA: Blue Crane Books, 1999), p. 26; and Masih
and Krikorian, Armenia: At the Crossroads, p. 7.
37. Malkasian, “Gha-Ra-Bagh!” p. 62.
38. Grisha Hairapetyan, Dashnak leader and paramilitary commander in Nagorno-Karabakh, inter-
view, Stepanakert, June 1999.
39. See Rena Pashabekova, ed., Bezhentsy (Baku: Gzhadzhlik, 1992), p. 425.
62
The Nagorno-Karabakh Conflict Revisited
Armenians suspected that Azeri refugees were being sent to Shusha by the
Azerbaijani authorities in a conscious effort to alter the demographic balance
in the province and thereby undermine the rationale for uniªcation with Ar-
menia. In protest, Armenians initiated strikes once again in Stepanakert in the
middle of September.40
One consequence of the intermingled ethnic settlement patterns in
Nagorno-Karabakh was that there were islands of Azeri settlements, such as
Shusha, within the predominantly Armenian-inhabited province. In some
places there were even small Armenian settlements within these Azeri en-
claves. The people living in the enclaves naturally began to worry about being
cut off from outside support during a conºict. Throughout the summer,
transports through the Lachin corridor had been hampered by civil distur-
bances, and the Armenian population in the NKAO began to fear an
Azerbaijani siege of the whole region, which would have been especially men-
acing given the harsh winters in the mountains.41 Similarly, a siege within a
siege threatened to develop around the Azeri village of Khojaly, slightly to the
north of Stepanakert. On 18 September 1988 a serious clash took place
among members of both ethnic groups, as well as Soviet MVD forces. One
Armenian was killed in a prolonged ªreªght. During the days that followed,
the ethnic cleansing of Stepanakert and Shusha was more or less completed.42
The clash in Khojaly was the most serious incident since the Sumgait ri-
ots. For the ªrst time, ªrearms were used on both sides. This meant that both
sides in Nagorno-Karabakh had the capacity and willingness to engage in a
semi-military confrontation. Moreover, the clash underscored the vulnerabil-
ity of the Armenian majority and the Azeri minority in the NKAO,
particulary to siege warfare. For the Armenians, Sumgait could no longer be
interpreted as an atypical, unintended consequence of hooliganism. Instead,
the Armenians concluded that the Azeris harbored hostility or even genocidal
intentions toward them.43
With the possible exception of Sumgait, the sporadic and relatively lim-
ited acts of violence described thus far did not necessarily indicate that any
central organization was directing the actions. Most of the clashes arose spon-
taneously and spread primarily because of the incompetence of local and cen-
tral authorities. But the subsequent waves of ethnic cleansing in both Arme-
nia and Azerbaijan would have been impossible without substantial planning
and centrally organized efforts. The entire Azerbaijani minority in Armenia
40. Malkasian, “Gha-Ra-Bagh!” p. 143. See also Croissant, Armenia-Azerbaijan Conºict, p. 31.
41. Malkasian, “Gha-Ra-Bagh!” p. 113.
42. Ibid., p. 146f; and Mityayev, “Karabakhskij Konºikt,” p. 490.
43. Rafael Kazaryan, member of the Karabakh Committee, interview, Erevan, June 1999.
63
Melander
44. Ibid.
45. Croissant, Armenia-Azerbaijan Conºict, p. 34.
46. Masih and Krikorian, Armenia: At the Crossroads, p. 13.
47. Zhanna Galstyan, interview, Stepanakert, June 1999.
64
The Nagorno-Karabakh Conflict Revisited
65
Melander
Azerbaijan. Soviet ofªcials most likely took this action because of growing dis-
satisfaction with the failed efforts to reconcile the hostile parties.49 Various
attempts at compromise had been met with hostility. Soviet MVD units
deployed in Nagorno-Karabakh encountered difªculty suppressing the guer-
rilla-style forces that were slowly developing with local support and that were
increasingly attacking the Soviet units to obtain weapons. These incidents had
caused the morale of the Soviet forces to decline. The situation ominously be-
gan to resemble the decade-long war in Afghanistan, from which the Soviet
military had only recently disentangled itself.50 (The last Soviet troops were
pulled out of Afghanistan in February 1989.) Moscow increasingly supported
Azerbaijan and took a stronger stance in favor of preserving the status quo and
the existing borders.
At ªrst, the 28 November decision did not have much effect on the situa-
tion. Armenians in the NKAO as well as in Armenia responded to the Su-
preme Soviet’s decision with another act of open deªance. On 1 December a
joint session of the Armenian Supreme Soviet and the national council of
Nagorno-Karabakh proclaimed a union of the Armenian republic and the
NKAO. In and around Nagorno-Karabakh Armenian guerrillas increased
their activities.51 On 9 January 1990 Armenian ofªcials completed a budget
for the Armenian republic that included funds for Nagorno-Karabakh for the
ªrst time. Azeri nationalists began to protest Moscow’s failure to enforce
Azerbaijani control over Nagorno-Karabakh.52 Rumors spread through
Azerbaijan that Armenian guerrillas were going to seize the province as well as
the strip of land separating it from the Armenian republic.53 On 11 January
Azeri militants, who reportedly were armed with military weaponry, attacked
three Armenian villages on the border with Nagorno-Karabakh and drove out
all the Armenians. Other militants purportedly representing the Azerbaijani
Popular Front declared that the Front had assumed power in the city of
Lenkoran and that the Communist party was dissolved.54 The ethnic violence
on the border of Nagorno-Karabakh soon spread to Baku, where rioters mur-
dered Armenians while the local police and the Soviet MVD troops in the city
stood idle. At least 74 people, most of them Armenians, were killed in Baku
during two days of chaotic riots.55
66
The Nagorno-Karabakh Conflict Revisited
On 19 January 1990, when the anti-Armenian riots in Baku had all but
ended, Moscow responded with a vehement military crackdown. Soviet
troops moved en masse into Baku and crushed the Azerbaijani Popular Front.
The Soviet Minister of Defense, General Dmitrii Yazov, openly admitted that
this was the primary goal of the operation.56 Outside the main towns, how-
ever, control was difªcult to reassert. Guerrilla warfare continued among na-
tionalist paramilitaries in the border regions between the NKAO and
Azerbaijan and between Armenia and Azerbaijan.
The central authorities in Moscow restored Soviet control over Baku and
installed a new Communist Party ªrst secretary in the Azerbaijani SSR, Ayaz
Mutalibov. Moscow then tried to quell the interethnic ªghting in the border
regions and to bolster the strength of the new Communist regime in
Azerbaijan, in part by providing military support to Mutalibov. Mutalibov, in
turn, steadfastly relied on Moscow and counted on the Soviet authorities to
resolve the Nagorno-Karabakh conºict favorably in exchange for Azerbaijan’s
loyalty.57 He attempted to disarm the paramilitaries of the Azerbaijani nation-
alist opposition and to reassert de facto control over Nagorno-Karabakh.58
Despite these efforts, the deadly riots in Baku had aggravated fears
among Armenians in Nagorno-Karabakh, who continued to feel threatened
by Azerbaijani rule.59 The failure of the Soviet MVD troops to intervene dur-
ing the January riots reminded the Armenians of their vulnerability and con-
vinced them that they needed their own military forces.60 As a result, the guer-
rilla bands fortiªed their ranks and the ªghting continued.
The escalation of the interethnic conºict after the abolition of direct rule
from Moscow in November 1989 followed the classic pattern of a spiral. To
the Armenians in Nagorno-Karabakh, the prospect that Baku would attempt
to enforce its rule over the province meant an increased danger, especially in
light of the widespread ethnic cleansings over the preceding year. By intensify-
ing the guerrilla campaign, the Armenian side could signal its resolve and abil-
ity to resist Azerbaijani rule with military means and thus try to deter the
leadership in Baku. The Azeri nationalist movement, on the other hand, was
increasingly dissatisªed with what it perceived as the government’s inaction
and lack of pressure on the breakaway province. By launching paramilitary
67
Melander
raids on Armenian settlements, the Azeri nationalists could signal their resolve
and intimidate the Armenians. The Azeri nationalists also wanted to send a
warning to the Communist regime in Baku.
The next step in the escalation of the Nagorno-Karabakh conºict came after
Operation Ring was initiated by the Soviet Army in cooperation with the
Azerbaijani special police in the spring of 1991. New evidence, including in-
terviews with participants, sheds valuable light on this hitherto underexplored
phase of the conºict.61 The account provided here has important implications
for the theoretical understanding of the escalatory process.
In early 1991 Soviet MVD forces came under ªre during serious clashes
between Azerbaijani special police and Armenian guerrillas. In early April
units of the Soviet Army’s 23rd Motorized Riºe Division were deployed as a
buffer along the Armenia-Azerbaijan border. By the end of the month, these
units had joined the MVD troops and the Azerbaijani special police in a mas-
sive offensive against Armenian settlements in Azerbaijan to the north of the
NKAO. Ofªcially, the purpose was to neutralize illegal guerrilla formations in
the area; but, in practice, Operation Ring amounted to systematic ethnic
cleansing. In May joint Soviet and Azerbaijani forces moved into Armenia
proper and arrested more than twenty people in three towns. They also at-
tacked Armenian villages in Nagorno-Karabakh. The brutality and scope of
the operation, and the fact that Soviet Army units worked openly with the
Azerbaijani forces for the ªrst time, had a tremendous impact on the Arme-
nian community.
Moscow launched Operation Ring to send a forceful signal to the Arme-
nians that their demands were unacceptable and that the price for continued
deªance would be high. According to the Armenian nationalist Zori Balayan,
General Yazov responded to the Armenian protests against Operation Ring
with an oft-repeated remark: “Well, Mutalibov and Ter-Petrosyan will just
have to sit down at a negotiating table.”62
On the previous occasion when Moscow had used large-scale military
force—the bloody assault on Baku in January 1990—Soviet troops had been
operating against the Azerbaijani nationalist opposition that threatened to
overthrow the rule of the Communist Party in that republic. Operation Ring
was the ªrst time that military force was directed on a large scale against the
61. The author carried out the interviews in 1997 and 1999.
62. Zori Balayan, Between Hell and Heaven: The Struggle for Karabakh (Erevan: Amaras, 1997), p. 60.
68
The Nagorno-Karabakh Conflict Revisited
Armenians. In earlier years Soviet leaders had wanted to avoid such a drastic
step. By the spring of 1991, however, Moscow had all but given up hope of
overcoming Armenian recalcitrance in the Nagorno-Karabakh conºict.
Weaker measures had repeatedly failed to convince Stepanakert to back down,
and the nationalist leadership in Erevan had signaled its unwillingness to sign
the new union treaty worked out by Gorbachev to prevent the break-up of the
Soviet Union.63
Operation Ring was also motivated by growing pressure from Azeri lead-
ers, who feared that the Armenian nationalist paramilitaries would create a
fait accompli by capturing the last Azeri-held settlements in Nagorno-
Karabakh and the strip of land separating the NKAO from Armenia. The au-
thorities in Baku sensed that they needed to strike at the Armenian guerrillas
and prevent additional Armenian forces and military supplies from entering
the enclave.64 The brutality with which the operation was carried out suggests
that the Azeris also wanted to intimidate the Armenians into submission or
ºight.
Although Operation Ring was not executed swiftly and decisively enough
to cripple the Armenian guerrillas, it did achieve some of its goals in the short
term.65 The massive damage inºicted on the Armenian population in the
summer of 1991 made the political leaders of Nagorno-Karabakh reconsider
the feasibility of their struggle. They signaled their willingness to retract most
of their demands once Operation Ring was halted. This move was not ini-
tially supported in Erevan, but after intense debate and the strong personal in-
tervention of the Armenian president, Levon Ter-Petrosyan (who even threat-
ened to resign), the parliament of the Armenian republic ofªcially endorsed
the initiative.66 Although a hardened and more militant Armenian leadership
eventually gained prominence in Stepanakert, the short-term effect of Opera-
tion Ring was to bring about concessions.
In late June a delegation was sent from Stepanakert via Erevan to Mos-
cow to seek assistance in launching negotiations with the Azerbaijani side.
The delegation met with several of the highest-ranking ofªcials of the Soviet
Union (but not with Gorbachev), and the initiative was generally welcomed.
Then, on 20 July, a delegation was sent from Stepanakert to Baku for direct
negotiations with Azerbaijani President Mutalibov.67 Political leaders in
63. Croissant, Armenia-Azerbaijan Conºict, p. 40f; and Herzig, “Armenia and the Armenians,” p. 261.
64. Musabekov, interview, Baku, June 1997.
65. Musabekov, interview 1997.
66. Souren Zolyan, member of the Karabakh Committee and the Armenian parliament, interview,
Erevan, June 1999.
67. Arutyunyan, Yanvar’ 1991 g.–Yanvar’ 1993 g.
69
Melander
68. The content of this letter can be found in Arutyunyan, Yanvar’ 1991 g.–Yanvar’ 1993 g., p. 54.
69. Mamed-zadeh, interview; and Georgi Petrosyan, member of the “National Council” and of the
team sent to Moscow in June 1991, interview, Stepanakert, June 1999; and Manvel Sarkisyan,
coordinator of Dashnak paramilitary preparations, interview, Erevan, June 1999.
70. Arutyunyan, Yanvar’ 1991 g.–Yanvar’ 1993 g., p. 54.
71. Karen Ohanjanian, member of Krunk and coordinator of the Stepanakert branch of the Helsinki
Citizens Assembly, interview, Stepanakert, June 1999; Petrosyan, interview; and Sarkisyan, interview.
72. See David Laitin, “National Revivals and Violence: Threats and Bluffs in East European Transi-
tions,” Archives Européennes de Sociologie, Vol. 36, No. 1 (1995), pp. 3–43.
70
The Nagorno-Karabakh Conflict Revisited
troops, and they no longer saw any reason to show restraint.73 Harsh criticism
of the meeting in Baku was broadcast over the radio on 29 July, stirring anger
and dismay among the more radical segments of the Armenian population of
Nagorno-Karabakh.74 That same day Armenian guerrillas from Nagorno-
Karabakh attacked and killed seven police ofªcers in a village previously cap-
tured by Azeri special police. The following day two more men were killed
and three taken hostage in another Armenian attack.75 These incidents may
well have been provocations aimed at undermining the prospects for further
negotiations. Then, on 10 August one of the Armenian participants in the
meeting in Baku, Valerii Grigoriyan, was murdered in a drive-by shooting in
the center of Stepanakert.76
This assassination of one of the old nomenklatura leaders in Stepanakert
was widely assumed to be a move by the radicals to silence the proponents of
negotiation with Baku. After this event the radicals became the dominant
force in Nagorno-Karabakh.77 According to ofªcials who were working for the
Azerbaijani government at the time, the leaders in Baku interpreted the inci-
dent in a similar manner.78 It is still unclear, however, who was in fact behind
the assassination. The Armenian militants did not admit responsibility, but a
statement by paramilitary commander Zhanna Galstyan indicates that, at the
least, there were militants who were willing to kill Grigoriyan. Asked about
the document that Grigoriyan and the others brought to Mutalibov, Galstyan
responded:
Anyone who signed such a document—a document that was invalid because
neither we nor the people could have abided by it—we would have threatened
his life; he would simply have been shot, even if this was a close friend of ours.79
The increased inºuence of this militant faction negated the prospects of a ne-
gotiated solution in the summer of 1991.
Another reason that the Armenians’ concessions in the summer of 1991
did not lead to full-ºedged peace was the abortive hardline coup in Moscow
in August 1991. The delegation sent by Stepanakert to Moscow in June did
not get very far, probably because most of the inºuential decision makers with
71
Melander
Conclusion
The main theoretical claim of this study is that in spite of the high levels of
willingness and opportunity present during the early stages of the escalatory
process, the Nagorno-Karabakh conºict would not have resulted in full-scale
war if the strategic uncertainties facing the actors had been overcome. This
claim can be supported by concrete counterfactual evidence drawn from dif-
ferent stages of the conºict. Neither the more general willingness-opportunity
framework nor the security dilemma provides a complete explanation for the
72
The Nagorno-Karabakh Conflict Revisited
war, since both theories offer an overly deterministic and pessimistic perspec-
tive on conºict. An analysis of the war must also recognize that decision mak-
ers based their actions on uncertain and conºicting expectations about the
consequences of alternative policies. Not all conºicts as intractable as
Nagorno-Karabakh necessarily result in all-out war; it should be possible to
ªnd other conºicts with similar levels of willingness and opportunity that
were resolved short of war.
In the case of Nagorno-Karabakh several courses of action could have
been pursued to avoid war. First, if the successive regimes in Azerbaijan and
their Azeri challengers had foreseen the ultimate dissolution of the Soviet
Union and the new Russian government’s shift in favor of Armenia, the Azeris
would reasonably have preferred to cut their losses. They could have pre-
vented the war by simply allowing Nagorno-Karabakh to secede. The regime
in Baku and the nationalist opposition chose to go to war against Armenia
over Nagorno-Karabakh because of their erroneous belief that Moscow (or
perhaps Ankara, in the minds of the nationalist opposition) would save the
day.
Second, if the relatively moderate Armenian leaders who were in power
in Nagorno-Karabakh until mid-1991 had been able to foresee the radicaliz-
ing effect that Operation Ring would have on the Armenian militants, they
would have had a strong incentive to retract their separatist demands before
the operation was initiated. The moderates would then have avoided the need
to present their failed capitulatory initiative, which cost them their leadership
positions (and in the case of Valerii Grigoriyan probably also his life). They
could have settled instead for the proposal advanced by Moscow in June 1988
to upgrade the status of the Nagorno-Karabakh Autonomous Province to an
autonomous republic within Azerbaijan.
Third, it is clear that the nomenklatura leaders of the Armenian separatist
movement in Nagorno-Karabakh were not willing to pursue their separatist
demands to the point of all-out war with the Soviet army. They stuck with
their program only so long as they believed that such a war was unlikely.
When Operation Ring in 1991 called this assumption into question, they
offered to retract their demands rather than ªght. It is reasonable to assume
that conºict could have been avoided if Moscow had sent stronger signals
earlier to the Armenians that their intransigence would be met with force.
Most likely, Gorbachev hesitated in using force until the last possible moment
because he feared that an open conºict would undermine his reform ambi-
tions. He probably also wanted the operation to appear as a response
to provocations by Armenian guerillas rather than as punishment for their
failure to abandon their demands for the uniªcation of the NKAO with Ar-
menia.
73
Melander
Fourth, if Moscow’s attention had not been diverted by the coup at-
tempt, the capitulatory initiative by the more moderate Armenian faction in
Nagorno-Karabakh might have resulted in an agreement between the Arme-
nians and Azerbaijanis to resolve the conºict short of war. The leadership in
Baku at this point was still very much under the inºuence of Moscow, as indi-
cated by Mutalibov’s close personal reliance on the second secretary of the
Communist Party of Azerbaijan, Viktor Polyanichko, who represented
Moscow’s interests in the republic.84 If Moscow had resolutely backed such an
agreement, the Azerbaijani government would have followed suit. The
Azerbaijani nationalist opposition would have had few reasons to undermine
such an agreement. Similarly, the government of Armenia supported the
capitulatory initiative. The militant Armenian faction in Nagorno-Karabakh
could not have pushed the conºict to full-scale warfare by itself if the other
main actors in the drama—the Armenian nomenklatura in Stepanakert, the
government of Armenia, the Soviet government, and the Azerbaijani govern-
ment, as well as its nationalist opposition—had all worked to prevent such an
escalation. According to an Armenian source in Stepanakert, the armed
groups in Nagorno-Karabakh had only 250 automatic riºes by mid-1991.85
This strongly suggests that as late as the summer of 1991 there was a very real
chance to avoid the escalation to full-scale war in Nagorno-Karabakh.
The pivotal importance of the militant faction in Stepanakert indicates
that Kaufman’s characterization of the Nagorno-Karabakh conºict as “mass-
led” rather than “elite-led” needs to be qualiªed. Although mass support for
the ªghting was strong, the moderates still might have been able to preserve
peace. The outcome of the struggle for primacy between the competing elites
at different levels made a crucial difference, and in this sense the conºict was
elite-led.
The fact that Operation Ring was successful in compelling the
nomenklatura leadership in Stepanakert to back down should not be taken to
mean that systematic ethnic cleansing is a legitimate or even very effective way
of dealing with ethnic separatism. It would have been preferable for both the
Armenians and the Azerbaijanis if the Armenians in Nagorno-Karabakh
could have been persuaded to accept autonomy within Azerbaijan in return
for credible guarantees that Azerbaijan would become a tolerant democracy in
which the Armenian minority could prosper. Unfortunately, the combination
of a history of tensions, an undemocratic and corrupt Soviet system, and a se-
curity dilemma prevented a peaceful resolution of the Nagorno-Karabakh
conºict.
74
The Nagorno-Karabakh Conflict Revisited
Note
This research was supported by grants from the Swedish Council for Research
in the Humanities and Social Sciences and the Swedish Agency for Civil
Emergency Planning. The views expressed here do not necessarily represent
those of the funding agencies or the Swedish Government. The author would
like to thank Mats Hammarström, Magnus Öberg, Svante Cornell, Kjell-Åke
Nordquist, and the anonymous reviewers for helpful comments.
75