Barbara A. Anderson Brian D. Silver

Population Redistribution and the Ethnic Balance in

No. 95-330

Research Report April 1995

Barbara A. Anderson is Professor of Sociology and Research Associate at the University of Michigan Population Studies Center, Ann Arbor, Michigan.

Brian D. Silver is Professor and Chair of the Department of Political Science at Michigan State University, East Lansing, Michigan.

Acknowledgments: The authors would like to thank Kathleen Dowley and Nate Silver for research assistance and Barry Stein for bibliographic materials. Research for this paper was supported in part by NICHD Grant Nos. RO1 HD-19915 and P30 HD-10003. The paper is planned for publication in Ronald Suny, ed., National and Social Change inTranscaucasia (Ann Arbor: University of Michigan Press). Abstract

Among the important long-term changes in the populations of Transcaucasia that give evidence of social relations among ethnic groups even during the period of Soviet control of the region was the reduction in the ethnic heterogeneity of the region and of the constituent republics (now countries). We examine evidence of these trends, both in terms of the outmigration of Russians and others from the region and in terms of the cross-migration of the major Transcaucasian nationalities before the massive flows of refugees began in the late 1980s. Three main factors led to this long-term reduction in ethnic heterogeneity in the Transcaucasus: high fertility rates of the indigenous populations, especially of the and ; emigration from Transcaucasia by members of non-Transcaucasian nationalities, primarily by Russians; and migration by members of Transcaucasian nationalities from neighboring Transcaucasian republics to their own republics. While these migration trends were probably driven in part by conditions in the overall labor market of the , they seem also to have been strongly affected by the social climate, local policies, and interethnic attitudes. In short, a trend toward indigenization was clear long before the upsurge of violence in the region. However, especially beginning in 1988 the pace of indigenization within Transcaucasia increased dramatically in connection with a communal conflict that resulted in major refugee flows.

Data sources: The primary data for this essay come from the Soviet censuses of 1926, 1959, 1970, 1979, and 1989. The titular nationalities of the three Transcaucasian republics together comprised only five percent of the population of the Soviet Union in 1989.1 This figure understates the importance of Transcaucasia to the Soviet Union, and it certainly does not reflect the importance of Transcaucasia to Russian and Soviet historical development. Transcaucasia has been less important because of the size of its population or its economic contribution than because of its strategic location and its linkages to and the Middle East.

The strategic location and historical ties of the Transcaucasus also help to justify our focus on population distribution and ethnic mixing. From a demographic perspective, numerical dominance is a vital factor in the relations among ethnic groups; it is both a consequence and a cause of ethnic antipathies and alliances.

Although it is common to observe that the dramatic political developments that led to the disintegration of the Soviet Union in 1991 were largely unanticipated at the beginning of perestroika in 1986, in the case of Transcaucasia there was evidence from basic long-term demographic trends to suggest the possibility of large changes and intensified ethnic conflict. In a book review published in the journal

Sotsiologicheskie issledovaniia [Sociological Studies] in January 1988, Ia. I. Rubin criticized Brian Silver for his explanation of the observed trends in migration in Transcaucasia between 1970 and 1979. Writing in an earlier edition of this monograph, Silver had concluded:

In the absence of a primary investigation of the motives of migrants, we can only speculate about the motives for the cross-migration of Armenians and Azeris. One plausible explanation is that the historic antipathy between members of the two groups has crystallized in recent years to encourage mutual avoidance and resettlement. Despite cultural policies in the Transcaucasian republics that have been aimed at reducing ethnic tension . . . an unfavorable cultural, administrative, or work environment for Armenians in and for Azeris in may have encouraged resettlement to their official homelands.

Alternatively, perhaps the cross-migration in Transcaucasia has another, less nationalistically tinged explanation. Namely, the very rapid rate of urbanization of Armenia in recent years, which has advanced that republic's level of urbanization ahead of the USSR as a whole, may have created significant opportunities for urban Armenians in Georgia and Azerbaijan to move out of those republics to Armenia. . . .2

1 Sodruzhestva Nezavisimykh Gosudarstv, Statisticheskii komitet, Itogi vsesoiuznoi perepisi naseleniia 1989 goda. Tom VII: Natsional'nyi sostav naseleniia SSSR [Results of the 1989 all-union census of population, Volume 7: Nationality composition of the Population of the USSR] (Moscow: 1993).

2 Brian D. Silver, "Population Redistribution and the Ethnic Balance in Transcaucasia," in Ronald G. Suny, Ed., Nationalism and Social Change in Transcaucasia (Ann Arbor, 1983): 277-278.

1 Rubin commented on Silver's argument as follows:

B. Silver sees the causes [of the resettlement of Armenians and Azerbaijanis from neighboring republics to their own republics] in "historically formed antipathies," in the still existing mutual hostility of Caucasian peoples, which "became aggravated to such a degree that they sought to flee from one another." The farfetched and tendentious character of such an explanation is partly exposed by the sovietologist himself. It is possible, he confesses, that the cause is the rapid urbanization of Armenia, the desire of people to take on an urban way of life.3

Rubin's comment was published in January 1988. The outbreak of violence between Azerbaijanis and

Armenians in Nagorno-Karabakh, , and elsewhere occurred in February 1988, and was followed in that and later years by a large-scale departure of Armenians from Azerbaijan and of Azerbaijanis from

Armenia.

Rubin's comments were made at a time when official doctrine of the Communist Party of the Soviet

Union had declared that fundamental antagonisms among nationalities had been liquidated, even if differences between nationalities remained. It was difficult to gather systematic information about popular and elite attitudes at that time. However, using demographic data we saw patterns of population change in the region that provided indirect evidence of antagonism. If this evidence from the period 1959-1979 could not have been used by itself to predict the outbreak of interethnic violence in 1988 and later, it nonetheless revealed a lot about fundamental relations between nationalities in Transcaucasia.

Among the important long-term changes in the populations of Transcaucasia that give evidence of social relations among ethnic groups even during the period of Soviet control of the region was the reduction in the ethnic heterogeneity of the region and of the constituent republics (now countries). We shall examine evidence of these trends, both in terms of the outmigration of Russians and others from the region and in terms of the cross-migration of the major Transcaucasian nationalities before the massive flows of refugees began in the late 1980s.

Three main factors led to this long-term reduction in ethnic heterogeneity in the Transcaucasus: high fertility rates of the indigenous populations, especially of the Armenians and Azerbaijanis; emigration from Transcaucasia by members of non-Transcaucasian nationalities, primarily by Russians; and migration by members of Transcaucasian nationalities from neighboring Transcaucasian republics to their own republics.

3 Ia. I. Rubin, book review in Sotsiologicheskie issledovaniia, No. 1 (January-February, 1988): 131-132.

2 While these migration trends were probably driven in part by conditions in the overall labor market of the

Soviet Union,4 they seem also to have been strongly affected by the social climate, local policies, and interethnic attitudes. In short, a trend toward indigenization was clear long before the upsurge of violence in the region. However, especially beginning in 1988 the pace of indigenization within Transcaucasia increased dramatically in connection with a communal conflict that resulted in major refugee flows.

Sources of Population Data: 1959–1989

The main data sources for this paper are the Soviet censuses of 1959, 1970, 1979, and 1989. Some data also are drawn from the 1926 census. Data related to changes since 1989 are more sparse and less systematic.

In our analysis for the period 1959-1989, we do not have direct information on migration patterns by ethnic group. Published data on migration from the former Soviet Union omitted cases of rural-rural migration. Moreover, very little information was published on migration by nationality. Hence, we can only infer net migration by comparing the population increase by nationality in a region with what might be expected based on the rate of increase in the population of that nationality in the Soviet Union as a whole.

Another factor to consider is that of emigration from the Soviet Union (including from the region of the former Soviet Union). We do not have information on the nationality of those who left by year or by place of origin in the Soviet Union. It is important to note, however, that the rate of emigration reflects not simply the factors (such as interethnic conflict) that may push people from their current location but also the factors that make it possible for them to move somewhere else, including legal factors, effective controls at borders, and the availability of work and other opportunities abroad. Thus, one should not read migration flows, even flows of refugees, as indicating only the effects of interethnic conflict.5

4 A. Vishnevsky and Zh. Zaionchskovskaia, Migratsiia iz SSSR: Chetvertaia volna [Migration from the Soviet Union: The Fourth Wave]. Russian Academy of Sciences, Center for Demography and Ecology of Man (Moscow, 1991); and "Internal and International Migrations in the Former Soviet Space," Presented at the Conference on Current and Emergent Migration within and from the Former USSR: Domestic and International Policy Perspectives, The Hague (March 4-5, 1993).

5 Flows of emigrants from the territory of former Soviet Union to the West since 1990 have probably been driven much more by perceptions of economic opportunity than by interethnic violence, despite the great publicity and seeming panic (continued...)

3 One must particularly keep in mind the mobility of the . After a period of repatriation to Armenia in the 1950s and 1960s, substantial emigration of Armenians from the Soviet Union took place in the 1970s and 1980s. There is evidence that disproportionate numbers of those who left, at least in the late 1970s and early 1980s, were "re-emigrants" who had previously repatriated to Armenia.6

A special word is needed concerning the conditions under which the 1989 census was conducted in Transcaucasia. The "critical date" of this census was January 12, 1989. A pogrom against Armenians in

Sumgait, leading to a reported 31 deaths, occurred in late-February, 1988, stimulated by the petition of the local soviet in Nagorno-Karabakh on February 20th for annexation to Armenia and a reported outbreak of violence against Azerbaijanis in , the capital of Nagorno-Karabakh; another outbreak of violence in November 1988 in and Kirovabad (Ganje) started a flow of Armenian refugees,7 who went in three

main directions: to Armenia (and in some cases later to Russia), to Russia, and to Karabakh (within

Azerbaijan). At the same time, many Azerbaijanis were driven from Nagorno-Karabakh, especially after armed hostilities grew and war broke out between Armenia and Azerbaijan. In November 1988, a reported

200,000 Azerbaijanis were expelled from Armenia and sent an enormous flow of refugees primarily to

(...continued) in Western European countries that the "Russians are coming." A sizeable literature appeared in the popular press, and a number of governments, including the European Community, commissioned special studies of this phenomenon. Despite headlines such as "Europe's Immigration Crisis," "Europe Braces for Immigrants, and "Communism's Failure Looses a Tide of Refugees," the "Russians" came only in moderate numbers, most were not, strictly speaking, "refugees," and most were not ethnic Russians. For an analysis of the characteristics and motivations of potential migrants from Ukraine, see Barbara A. Anderson, Edward Ponarin, Brian D. Silver, and Mikk Titma, "Emigration Potential Among Young Adults in Ukraine," Rand Corporation Report No. DRU-567-FF, for the Project on Migration within and from the Former Soviet Union (November, 1993).

6 In our work on the sample for the Soviet Interview Project, we studied the backgrounds of 2,000 Armenian refugees from the Soviet Union who had received assistance through the Tolstoy Foundation between 1979 and 1982. Of this group, 60% had not been born in the USSR, and most of the others were members of their families. Thus, most of these Armenians were re-emigrants who had formerly repatriated to Armenia from abroad. Of course, many more Armenians left the USSR under other auspices, so those assisted by the Tolstoy Foundation may not have been representative. It is also possible that many of these individuals did not hold Soviet passports. See Barbara A. Anderson and Brian D. Silver, "Descriptive Statistics for the Sampling Frame Population: The Eligible Population for the Soviet Interview Project General Survey," University of Illinois, Soviet Interview Project Working Papers, No. 2 (1986).

7 Elizabeth Fuller, "What Lies behind the Current Armenian-Azerbaidzhani Tensions?" Report on the USSR, May 24, 1991: 12-15.

4 Azerbaijan. On December 7, 1988, Armenia was struck by a major earthquake that cost 25,000 lives and left half a million people homeless.

For all of these reasons, when the census results for Armenia and Azerbaijan were first released in the summer of 1989, they were described as "tentative." Even the "final" results must be treated with reserve.

Because the census was supposed to count the population according to their "permanent" residence,8 it is not at all clear how this residence was determined for persons who were living in temporary quarters because of the earthquake or because they had been driven from their permanent home by war or civil strife.

Armenia and Azerbaijan are not the only countries in which civil conflict affected the numerical counts of population in 1989. Ethnic conflict joined with what might be called "census politics" in Georgia.

Although the civil war between Abkhazians and Georgians had not yet broken out at the time of the 1989 census, there had long been hostility between these nationalities and in summer of 1988 a group of Abkhazian communists sent a letter of grievances against the Georgians to the Nineteenth Party Conference in Moscow.9

Mass rallies in Abkhazia did not begin until March 1989, and the open escalation of political and later military hostilities did not come until later. Nonetheless, suspicions about the accuracy of the count of persons by nationality led to special precautionary measures. When the census forms were completed in Abkhazia in

January 1989, members of an "Abkhazian society" and a "Georgian society" played the roll of "poll watchers" by observing the counting of the forms both in Abkhazia and in Tbilisi.10

Similarly, hostility between Georgians and Ossetians had long existed but also did not manifest itself at a mass level before the 1989 census or take a sharp turn for the worse until the Georgian government abolished South Ossetia's autonomous province status in December 1990, precipitating open guerilla warfare and a flow of refugees (and deportees) from Ossetia. However, harassment of Ossetians by Georgians, which

8 Barbara A. Anderson and Brian D. Silver, "`Permanent' and `Present' Populations in Soviet Statistics," Soviet Studies 37 (July, 1985): 386-402.

9 Elizabeth Fuller, "Georgia, Abkhazia, and Checheno-Ingushetia," RFE/RL Research Report, February 7, 1992: 3-7.

10 Personal conversation by the authors with census officials in Georgia in November 1990.

5 by mid-1992 had driven more than 100,000 refugees from South Ossetia to Russia (mainly to North Ossetia),11 may have affected the validity of the census count of Ossetians in Georgia in 1989.

Needless to say, such antagonisms could have affected the accuracy of counts by nationality in earlier censuses as well. There is evidence from an examination of intercensal changes in the number of

Georgians in earlier censuses to suggest that the number of Georgians was growing at a rate that cannot readily be accounted for by natural increase. Hence, either there was a process of assimilation or there was a process of systematic miscounting of persons by nationality in the census. Whether this was at the expense of

Ossetians, Abkhazians, or some other nationalities, cannot readily be determined.12

Hence, we must treat the census figures by nationality from the region with some caution, especially in 1989. To be sure, these are not the only questions that have been raised about the accuracy of the counts by nationality.13 However, for many of the smaller nationalities, the census count in 1989 appears to have been more complete than previous census counts. Clear examples in Transcaucasia are the Georgian Jews and the Talysh (who lived mainly in Azerbaijan).14 The period around the time of the census was a time of flourishing and even a euphoria of national consciousness in many regions of the Soviet Union, which may have reduced the pressure on smaller nationalities to appear to assimilate to larger ones.

This also affected some of the more sizeable nationalities. For example, it appears that there was a more complete count of Tajiks in Uzbekistan in 1989 than in 1979.15 The number of Ukrainians counted in

11 Rozalina Ryvkina, Rostislav Turovskiy, Robert J. Brym, and Patricia Patchet-Golubev, The Refugee Crisis in Russia (Toronto: York Lanes Press, 1992).

12 Barbara A. Anderson and Brian D. Silver, "Estimating Russification of Ethnic Identity Among Non-Russians in the USSR," Demography 20 (November, 1993): 461-489.

13 Brian D. Silver, "The Ethnic and Language Dimensions in Russian and Soviet Censuses," in Ralph S. Clem, Ed., Research Guide to the Russian and Soviet Censuses (Ithaca: Cornell University Press, 1986): 70-97; and Barbara A. Anderson and Brian D. Silver, "Demographic Sources of the Changing Ethnic Composition of the Soviet Union," Population and Development Review 15 (December, 1989): 609-656.

14 For further discussion, see Anderson and Silver, "Demographic Sources." In the case of the Talysh, what other than a change in the overall political context of the census can account for the apparent increase in their numbers from 1,400 in 1979 to 22,000 in 1989? Similarly, so-called (mainly in Daghestan) increased from 9,400 to 19,000 and Georgian Jews from 8,500 to 16,000. Crimean increased from 132,000 to 272,000 in the USSR.

15 Anderson and Silver, "Demographic Sources."

6 the 1989 census also exceeded the expected number by several hundred thousand persons.16 Even before this era, however, there was evidence from many nationalities in the former Soviet Union that self-identified

"nationality" could shift within a person's lifetime, even though official documents labeled the individual more or less permanently "by nationality" from age 16.17 Given that nationality in Soviet censuses was determined primarily according to the individual's self-identification,18 the distribution of the population by nationality could change in response to perceived opportunities and incentives for persons to signify their identity with particular ethnic labels. How else to explain why the Talysh, an Iranic people who had long been undergoing assimilation by the Azerbaidzhanis, suddenly increased more than ten-fold between the 1979 and 1989 censuses?

Population Growth and Distribution: 1959–1989

Table 1 shows the number of members of the three titular Transcaucasian nationalities (Armenians,

Azerbaijanis, and Georgians) and of Russians in the USSR as a whole and in each Transcaucasian republic at the time of each post-War Soviet census. It also shows the percentage by which each group increased or decreased between censuses.

We first examine the top panel of the table, in which the population size and percentage growth of each major Transcaucasian nationality and of Russians in the USSR as a whole is shown. In every intercensal period, the Transcaucasian nationalities exhibited a higher rate of growth than Russians in the USSR as a whole, mainly due to their higher fertility.19 Given the growth rates of these nationalities in the USSR as a whole, if they continued to grow at the same pace as they did between 1979 and 1989, the number of

16 A. V. Topilin, "Vliianie migratsii na etnonatsional'noi srede [The Influence of migration on ethno-national composition]," Sotsiologicheskie issledovaniia, 1992, No. 7: 31-42.

17 Anderson and Silver, "Estimating Russification"; Barbara A. Anderson and Brian D. Silver, "Demographic Consequences of World War II on the Non-Russian Nationalities of the USSR," in Susan J. Linz, Ed., The Impact of World War II on the Soviet Union (Totowa, NJ: Rowman and Allanheld, 1985): 207-242; and Barbara A. Anderson and Brian D. Silver, "Some Factors in the Linguistic and Ethnic Russification of Soviet Nationalities: Is Everyone Becoming Russian?" in Mark Beissinger and Lubko Hajda, Eds., The Nationality Factor in Soviet Politics and Society (Boulder, CO: Westview, 1990): 95-130.

18 Silver, "The Ethnic and Language Dimensions."

19 Anderson and Silver, "Demographic Sources."

7 Azerbaijanis would double in 33 years, the number of Georgians and Armenians would double in 64 years, and the number of Russians would double in 126 years.20 Thus, the number of Azerbaijanis was growing about twice as fast as the number of Armenians and Georgians, and Armenians and Georgians about twice as fast as Russians.

[Table 1 about here]

The pace of growth of all four nationalities slowed considerably over time. This was mainly due to declines in fertility. Russians had the lowest fertility of all four groups in every intercensal period, and

Azerbaijanis the highest. In the 1960s and 1970s, fertility of Armenians was higher than that of Georgians, but by the 1980s, the fertility of Georgians and Armenians was comparable.

Part of the increase in the number of Armenians was due to repatriation of Armenians from abroad.

Between 1946 and 1975, approximately 150,000 Armenians repatriated to the USSR and almost all of them settled in Armenia. Of these, 90,000 arrived in 1946-1948; the remainder came later.21 During the 1960s, an average of 2,000 to 4,000 Armenians per year repatriated to Armenia from abroad.22

Table 1 contains strong evidence of migration of Russians out of Transcaucasia. In Georgia,

Russians declined in absolute numbers in every intercensal period since 1959; in Azerbaijan, the number of

Russians declined since 1970; and in all three Transcaucasian republics, the number of Russians declined between 1979 and 1989. This decline is probably due almost entirely to outmigration of Russians, not to changes in mortality and fertility or to assimilation of Russians by the local nationalities.23

Assuming that fertility and mortality of a nationality was similar among those who lived outside and those who lived inside the borders of a given republic, then the growth in the number of members of a nationality within the given republic would be a result of both the growth rate of that group as a whole and

20 For further discussion and illustrations of this approach see Anderson and Silver, "Estimating Russification."

21 V. A. Kuregian, "Nekotorye osobennosti formirovaniia i ispol'zovaniia trudovykh resursov Armianskoi SSR [Some peculiarities of the formation and utilization of labor resources in the Armenian SSR]," in V. G. Kostakov and Ie. L. Manevich, Eds., Regional'nye problemy naseleniia i trudovye resursy SSSR (Moscow: Statistika, 1978): 237-254.

22 Bol'shaia sovetskaia entsiklopedia [Large Soviet Encyclopedia], 3rd Ed.,Vol II (Moscow: Izdatel'stvo Sovetskoi entsiklopedii, 1970): 643.

23 Vishnevsky and Zaionchkovskaia, Migratsiia iz SSSR.

8 net migration across the boundaries of the republic.24 Comparison of the percentage by which members of a nationality grew in a republic and in the USSR as a whole indicates whether on balance members of that nationality migrated into or out of that republic.25

Even though the number of Russians in Armenia increased between 1970 and 1979 (by 5.9%) this was less than the percentage increase of Russians in the USSR as a whole (6.5%). Similarly, although the number of Russians in Azerbaijan increased between 1959 and 1970 (by 1.8%) this was far less than the percentage increase of Russians in the USSR as a whole in that period (13.1%). The information in Table 1 suggests that only in Armenia in the period 1959-1970 did Russians show net immigration to a Transcaucasian republic.

Table 1 shows that the growth of the number of Armenians in Armenia was far higher than the growth in the number of Armenians in the USSR as a whole in every intercensal period. For 1959-1970 and for some of 1970-1979, repatriation of Armenians from abroad was part of the reason for the large increase of Armenians in Armenia. However, migration by Armenians to Armenia from elsewhere in the Soviet Union has probably been the main source of the large increase in the number of Armenians in Armenia since 1959.

This interpretation is also supported by the evidence of a decline in the number of and in Georgia in both 1970-1979 and 1979-89.

The number of Azerbaijanis in Azerbaijan increased at about the same pace as the number of

Azerbaijanis in the USSR as a whole. However, in Armenia and in Georgia, the percentage growth of

Azerbaijanis in every intercensal period was less than that of Azerbaijanis in Azerbaijan, which implies net outmigration of Azerbaijanis from those regions. Not surprisingly, there was a large drop in the number of

Azerbaijanis in Armenia between 1979 and 1989, probably mostly in connection with the mass expulsion of

24 Strictly speaking, this statement also depends on the propensity of different nationalities (and of persons from the same nationality living in different republics) to emigrate abroad. But this is not the most important factor affecting the changing ethnic composition of the Transcaucasian republics.

25 For further discussion and applications of this approach, see Anderson and Silver, "Demographic Sources." Topilin ("Vliianie migratsii") appears to have adopted a similar approach and reached similar conclusions, but he does not explain his methodology.

9 Azerbaijanis from Armenia in late 1988.26 However, even before the mass exodus beginning in late 1988,

Azerbaijanis appear on balance to have been moving out of Georgia and Armenia.

The number of Georgians in Georgia increased at about the same pace as that of Georgians in the

USSR as a whole. However, there was not much potential for Georgians to move to Georgia, since over 95% of all Georgians in the USSR lived in Georgia at every post-War census date. The number of Georgians living in Armenia remained very small throughout the 1959-1989 period.

Table 2 shows the distribution of Armenians, Azerbaijanis, and Georgians among the

Transcaucasian republics, Russia, and the rest of the USSR, at the dates of the four post-War Soviet censuses.

Thus, while Table 1 compared the growth of the population of various nationalities over time, Table 2 considers of geographic distribution of members of the various nationalities at given dates. All of these groups have been concentrated residentially in the Transcaucasus, especially in their titular republics. At every census date, it was very rare for Georgians to reside outside of Georgia. Those Georgians who did not live in Georgia tended to be located in Russia. Azerbaijanis were somewhat more likely than Georgians to live outside of their titular republic. But with each successive census, the chance that an Azerbaijani lived in

Armenia or Georgia declined, while the proportion of Azerbaijanis residing in Russia increased.

Armenians were much more widely dispersed than Azerbaijanis or Georgians, both within

Transcaucasia and elsewhere in the USSR. However, the proportion of Armenians who resided in Armenia increased at every census date. Also, the proportion of Armenians who lived in Azerbaijan or Georgia

26 The widely reported figure of more 200 thousand Azerbaijani expellees from Armenia in November 1988 seems to assume that all Azerbaijanis were expelled from Armenia at once. At first glance, the figure appears to be inconsistent with the census data, which indicated that there were 160.8 thousand Azerbaijanis in Armenia in 1979. But if one starts with the figure of 160.8 thousand Azerbaijanis living in Armenia at the time of the 1979 census and if this population grew 23.6% by natural increase between 1979 and 1989 (the rate of increase in the number of Azerbaijanis in the USSR as a whole), then the number of Azerbaijanis in Armenia would have been about 200,000; in fact, given that the Azerbaijanis in Armenia were predominantly rural, the rate of natural increase was probably significantly higher than the figure assumed in this calculation. However, the 1989 census reported that 84.9 thousand Azerbaijanis still resided in Armenia at the time of the census (after the exodus had occurred). This number is not consistent with the implication that all Azerbaijanis were expelled from Armenia at once. However, this process may actually have been completed only after the January 12, 1989 census date. Moreover, as noted previously, there had to be substantial difficulty in establishing what the "permanent" population of Armenia and Azerbaijan was at the time of the 1989 census, including implementing the normal census procedures for attributing "temporary" residents to their permanent place of residence. We do not know how these problems were dealt with by the census administrators.

10 declined at each census date. Armenians also showed an overall decrease in the proportion living in Russia between 1959 and 1979, but this reversed between 1979 and 1989, after the major refugee flows began.

[Table 2 About Here]

While the concentration of Transcaucasian nationalities in their traditional homelands increased fairly steadily between 1959 and 1989, at least as striking in the 1979 to 1989 period is the movement of

Azerbaijanis and Armenians away from parts of the Transcaucasus that are not their traditional homelands, often to Russia. This redistribution to Russia was probably primarily a result of the ethnic strife in Azerbaijan in February 1988 and the of Azerbaijanis from Armenia in November 1988. To be sure, the majority of refugees from Azerbaijan and Armenia did not go to Russia, but substantial numbers did go there.

Those Armenians from Azerbaijan who were more likely to take refuge in Russia than in Armenia were reportedly the ones who were russified or married to non-Armenians (e.g., to Russians, Ukrainians, or

Azerbaijanis).27

Table 3 shows the percentage of the population living in urban areas at each Soviet census date for each Transcaucasian republic and for the USSR as a whole. It also shows the percentage of the titular nationality living in urban areas and the percentage of Russians living in urban areas in each republic.

[Table 3 About Here]

The very rapid rate of urbanization of Armenia since the 1950s may have created significant opportunities for urban Armenians residing in Georgia and Azerbaijan to move out of those republics to

Armenia. Note that in 1959, 55% of Armenians in Georgia and 61% of Armenians in Azerbaijan lived in urban areas – well ahead of the average for each of those republics. At the same time, those rural Armenians in Georgia and Azerbaijan who decided to move to an urban location could more readily find work in the fast- developing urban Armenia than in the more slowly urbanizing Georgia and Azerbaijan.

27 U. S. Committee for Refugees (USCR), Faultlines of Nationality Conflict: Refugees and Displaced Persons from Armenia and Azerbaijan (Washington, D.C.: U.S. Committee for Refugees, 1994).

11 Ethnic Dominance

The impact of both natural and mechanical increases in population on the ethnic homogeneity of the Transcaucasian republics is apparent in Table 4, which reveals that in each republic between 1959 and

1989, the titular nationality increased its share of population in both urban and rural areas in every intercensal period except for some slight fluctuations in rural Armenia. In contrast, Russians – the largest non-

Transcaucasian nationality in the region – declined as a proportion of the population.

[Table 4 About Here]

The change is particularly striking in Azerbaijan. In both the rural and urban populations,

Azerbaijanis have come clearly to dominate numerically; Russians, always a minuscule proportion of the rural population of Azerbaijan, in 1989 comprised only one-tenth of the urban population. In short, by 1989

Azerbaijan had lost almost all of its Russian population. In none of the three Transcaucasian countries today do Russians comprise more than 6% of the population.

Thus, the high rates of natural increase of the Transcaucasian nationalities (especially of Armenians and Azerbaijanis), the cross-migration into their titular republics (states) by Armenians and Azerbaijanis, and the departure of Russians (especially from Azerbaijan and Georgia) have combined to reduce ethnic intermingling. Not only are all of the Transcaucasian states clearly dominated numerically by the titular nationality, but so are the urban and rural areas and the leading industrial and political centers. Even in Baku, where Russians and other non-Azerbaijanis played a major role in industrial (not to mention revolutionary) development, Russians fell to a numerical minority in the 1960s and have lost ground since then. Of course, in the opinion of many Armenians, Azerbaijanis, and Georgians, the indigenization is not yet complete. The conflict over Nagorno-Karabakh, with its primarily Armenian population living within Azerbaijan, continues.

Moreover, remaining members of minority nationalities, including Abkhazians and Ossetians in Georgia, and less sizeable nationalities in the other countries, remain in a difficult circumstances.

Intermarriage Rates

Ethnic intermarriage rates are often examined as an indicator of antagonism or closeness among ethnic groups. The possibilities for ethnic intermarriage are obviously constrained by the levels and patterns

12 of ethnic group mixing in a region or social setting. Quite apart from their attitudes toward other nationalities, few Azerbaijanis should be expected to marry non-Azerbaijanis, for example, because Azerbaijanis are so heavily concentrated in their own republic and because Azerbaijanis constitute a large (and increasingly larger) proportion of the population of the republic.

Soviet sociologists developed techniques for taking into account the degree of ethnic mixing when measuring preferences for ethnic endogamy. Essentially, the technique measures the attraction or repulsion between nationalities against a standard random selection of marriage partners: when the number of marriage partners selected from a particular nationality exceeds or falls short of what one would expect if spouses were selected at random from the population of potential spouses, one can speak of a special attraction or aversion to marriage between particular nationalities. According to one such measure reported by the Ukrainian demographer L. V. Chuiko, based on marriages that took place in the fourteen non-Russian republics in 1969, the titular nationalities of these republics scored as follows (from the highest to the lowest) in their preference for ethnic endogamy:28

Kyrgyz 95.4 Tajiks 77.3 93.6 Lithuanians 68.2 90.7 Moldovans 62.0 Azerbaijanis 89.8 Latvians 61.4 86.2 Belarussians 39.0 Georgians 80.5 Ukrainians 34.3 Estonians 78.8 Armenians 33.4

The scale runs from +100 (highest possible preference for ethnic endogamy) to -100 (highest preference for ethnic exogamy), with 0 representing indifference to the nationality of the spouse. According to this measure, both Azerbaijanis and Georgians displayed a high preference for ethnically homogeneous marriages, while the Armenians had the lowest preference for ethnic endogamy of any union republic nationality. Armenians thus appeared to be rather susceptible to assimilation through intermarriage.

Nevertheless, because of the extremely high ethnic homogeneity of Armenia, the actual proportion of ethnic intermarriages taking place in Armenia was probably much lower than in most other republics of the

Soviet Union. Indeed, in 1970 only 3.7% of all married couples in Armenia were ethnically mixed, much

28 L. V. Chuiko, Braki i razvody [Marriages and Divorces] (Moscow, 1975): 69.

13 lower than the proportion of mixed marriages in any other union republic and than the USSR-wide figure of

13.5%. (At the same time, Azerbaijan ranked second from the bottom and Georgia fourth from the bottom among union republic nationalities on the same measure.)

Of course, one must be wary of such aggregate figures, because they do not tell us which nationalities in each republic are intermarrying – for example, Kazakhstan ranked second among the union republics in 1970 in the proportion of mixed marriages in the population; but we know from other evidence that the overwhelming majority of mixed marriages concluded in Kazakhstan were concluded between members of different Slavic nationalities (namely, Ukrainians and Russians), not between Kazakhs and

Slavs.29 In any case, the extreme ethnic homogeneity of Armenia means that the Armenians's appearance of openness (lack of prejudice) regarding mixed marriage resulted in very few actual mixed marriages within

Armenia. The situation for the Armenian diaspora, on the other hand, was probably quite different.

This conclusion finds support also in a study by A. E. Ter-Sarkisiants, which examined intermarriage rates in Armenia for the years 1967, 1969 and 1970.30 Between 93 and 98 percent of marriages concluded in Armenia in 1967 to 1970 were between spouses of the same nationality. Four-fifths of the mixed marriages occurred in urban areas of the republic. The most common type of mixed marriage was between an Armenian man and a Russian woman (or a member of another non-Transcaucasian nationality). By contrast, Armenian women rarely married non-Armenian men. Furthermore, marriages between Armenians and Azerbaijanis were much less frequent than marriages between Armenians and Russians. Finally,

Armenians displayed a strong preference for ethnic endogamy overall (perhaps in contradiction to Chuiko's data).

29 A. P. Iegurnev, "Mezhnatsional'nye braki i ikh rol' v sblizhenii natsii i narodnostei SSSR [Interethnic marriages and their role in the rapprochement of nations and peoples in the USSR]," Nauchnyi kommunizm, 1973, No. 4: 28-34; Iu. A. Ievstigneev, "Mezhetnicheskie braki v nekotorykh gorodakh severnogo Kazakhstana [Interethnic marriages in some cities of northern Kazakhstan]," Vestnik Moskovskogo universiteta,Seriia istorii, 1972, No. 6: 73-82.

30 A. E. Ter-Sarkisiants, "O natsional'nom aspekte brakov v Armianskoi SSR [On the nationality dimension of marriages in Armenia]," Sovetskaia etnografiia, 1973, No. 4.

14 Language as a Measure of Ethnic Accommodation

Language use represents another measure of the strength of ethnic attachments and of interethnic group accommodation. The Soviet regime long sought to spread Russian as a lingua franca (the so-called

"language of internationality communication") among the non-Russian nationalities. Efforts to improve and expand the Russian-language curriculum were particularly intense in the 1970s, spurred perhaps both by the

1970 census report of a low level of knowledge of Russian as a second language among many non-Russian nationalities and by the regime's concern about the large southerly regional shift in the source of new recruits into the civilian and military manpower pools.31

Soviet leaders also had long been committed in both theory and practice to conducting basic instruction in schools in the native languages of non-Russian nationalities. This policy of encouraging the study of and conducting basic instruction in the national languages was especially characteristic of the fourteen titular nationalities of the non-Russian union republics, where instruction in most subjects was conducted in the native language for most indigenous non-Russian pupils through secondary school and often into higher education, while Russian was taught as a separate subject in school, typically from the first or second year.32

The results of Soviet language policy were mixed. The most obvious shortcoming (from the regime's perspective) was the low levels of knowledge of Russian that were achieved, especially among many of the larger non-Russian nationalities. Another shortcoming (not necessarily from the regime's perspective) was that among some nationalities, especially the smaller ones and the non-Muslim ones, Russian language was gradually becoming not a second language but a primary one.33

31 Brian D. Silver, "The Status of National Minority Languages in Soviet Education: An Assessment of Recent Changes," Soviet Studies 26 (January, 1974): 28-40; Barbara A. Anderson and Brian D. Silver, "Equality, Efficiency, and Politics in Soviet Bilingual Education Policy: 1934-1980," American Political Science Review 78 (December, 1984): 1019-1039.

32 Anderson and Silver, "Equality and Efficiency," and "Some Factors." Even for the large, so-called "union republic nationalities," this policy was uneven, however. In Belarus in the last two decades of Soviet rule there were no urban Belarus-language secondary schools. In Ukraine, the availability of Ukrainian-language secondary schools varied by province and in many urban areas only Russian schools were available.

33 Brian D. Silver, "Language Policy and the Linguistic Russification of Soviet Nationalities," in Jeremy R. Azrael, (continued...)

15 The language behavior of the Armenians, Azerbaijanis, and Georgians exemplifies the first

"shortcoming": fluency in Russian was limited in scope, while the traditional national tongues were thriving.

Although, as we shall see, in the capital cities of Azerbaijan and Georgia, there were many russified

Armenians, on the whole the members of the major Transcaucasian nationalities were not very russified linguistically, if the main criterion is the designation of the native language rather than the learning of Russian as a second language.

An understanding of Soviet language policy makes it clear why the Transcaucasian nationalities were able to preserve their native tongues. In the Soviet context, the major Transcaucasian languages have been favored: Armenian and Georgian use their own alphabets; for all three nationalities, institutions of secondary and higher education were abundant. Not all Trancaucasians in the Soviet period had the opportunity to attend or chose to attend native-language schools. But in comparison with the languages of such other union republic nationalities whose official homelands ranked lower in the federal hierarchy, the major Transcaucasian languages were treated well.34

The impact of Soviet language policy cannot be understood in isolation from the demographic and cultural settings in which it occurred.35 That bilingualism (knowledge of Russian as a second language) was not widespread among the major Transcaucasian nationalities must be attributed in part to the low levels of interethnic mixing in Transcaucasia. The increasing ethnic homogeneity of Armenia and Azerbaijan, as well as the continued limited presence of Russians in all three Transcaucasian republics, helped to retard the spread of Russian as a second language.

(...continued) Ed., Soviet Nationality Policies and Practices (New York: Praeger, 1978): 250-306.

34 For comparisons of native-language schooling among the fourteen non-Russian union republic nationalities, see Jonathan Pool, "Soviet Language Planning: Goals, Results, Options," in Azrael, Soviet Nationality Policies and Practices: 230-231. For a very low estimate of provision of native-language schools in Belarus, in contrast with Armenia, see K. Kh. Khanazarov, Reshenie natsional'no-iazykovoi problemy v SSSR [Solution to ethnic-linguistic problems in the USSR] (Moscow: Politizdat, 1977). For figures on Azerbaijanis and Armenians in Azerbaijan, see A. N. Baskakov, "O funktsionirovanii russkogo iazyka v Azerbaidzhanskoi SSR [On the functioning of Russian language in Azerbaijan]," in F. P. Filin et al., Eds., Russkii iazyk kak sredstvo mezhnatsional'nogo obscheniia [Russian language as a means of inter- nationality communication] (Moscow).

35 Silver, "Language Policy and Linguistic Russification."

16 Perhaps because language and educational planners could not do anything about the larger social or ethnic settings within which pupils are located, they tended to ignore the special opportunities or problems afforded by the ethnic setting. Even language surveys gave only a partial picture of the importance of social context for the growth of bilingualism. For example, in a large-scale multi-republic survey of language practices, respondents were asked, "Where did you learn your second language?" Majorities of the titular nationalities of Moldova, Estonia, Uzbekistan, and Georgia named "school" far more frequently than any other source (alternative sources given were "army," "family," "friends," and "higher educational institution").36 The younger the respondent, the more frequently "school" was named as the leading source of the learning of

Russian. One should scarcely be surprised that school was named most frequently as the source of learning of Russian. It was in schools that non-Russians usually first engaged in the formal study of Russian grammar, literature and culture.

But gaining fluency in Russian was probably difficult when school was virtually the only source of learning Russian. It is clear from writings of the distinguished linguist A. N. Baskakov, for example

(although Baskakov does not emphasize it), that the difficulty that Azerbaijanis had in learning to pronounce

Russian words correctly, let alone to speak the language grammatically correctly, was linked to the limited contact between Azerbaijanis and Russians speakers outside the classroom.37 Without intensive interethnic contact, teaching Russian was more akin to teaching a foreign language than it is to teaching a "second mother tongue" (as Soviet officials sometimes labelled the Russian language), even if exposure to Russian-language mass media was pervasive.

We have argued elsewhere that shift of native language, as opposed to learning of Russian as a second language, usually denoted a serious change in ethnic identity.38 In a majority of cases, non-Russians

36 S. I. Bruk and M. N. Guboglo, "Faktory rasprostraneniia dvuiazychiia u narodov SSSR (po materialam etnosotsiologicheskikh issledovanii) [Factors in the spread of bilingualism among peoples of the USSR (According to materials from ethnolinguistic studies)]," Sovetskaia etnografiia, 1975, No. 5: 17-30; M. N. Guboglo, "K izucheniiu perspektive razvitiia duvuiazychiia u narodov SSSR [Toward study of the prospects for the development of bilingualism among peoples of the USSR]," Istoriia SSSR, 1978, No. 1: 27-42.

37 Baskakov, "O funktsionirovanii russkogo iazyka."

38 Brian D. Silver, "Methods of Deriving Data on Bilingualism from the 1970 Soviet Census," Soviet Studies 27 (continued...)

17 who claimed Russian as their native language lost command of their own nationality's language as a second language. Moreover, adoption of Russian as native language meant that the person was highly russified, even if they did not switch their ethnic self-label to Russian.39 On the other hand, learning of Russian as a second language (the spread of bilingualism) did not mean much in terms of ethnic identity; it was most often a pragmatic adjustment, determined by both needs and opportunities to learn Russian.

As the data in Table 5 reveal, the knowledge of Russian as a second language among Armenians,

Azerbaijanis and Georgians residing in their own republics was rather low (see columns 4, 5, and 6). Shift of native language to Russian was especially limited. On the other hand, both knowledge of Russian as a second language and shift of mother tongue to Russian occurred much more frequently among Transcaucasians who resided outside their official homelands. In both 1959 and 1970, for example, over one-fourth of the

Georgians who resided outside Georgia claimed Russian as a native language. The corresponding percentages among Azerbaijanis and Armenians were lower, but the contrast in the knowledge of Russian between those

Transcaucasians who resided inside their own republics and those who resided outside is quite marked for all three nationalities.

[Table 5 About Here]

There were sharp differences in linguistic russification between urban and rural residents. Among residents of their official republics, urban-rural differences in the adaption of Russian as a native language were quite small in percentage terms for all three Transcaucasian groups (Table 5). But urban-rural differences in the knowledge of Russian as a second language were large. The limited shift to Russian as a native language, even in urban areas, attests to the strength of the ethnic identities of these nationalities and is consistent with the evidence of limited inter-ethnic marriage.

Although urban settings were more conductive to russification than rural settings, in Soviet

Transcaucasia they were not fertile ground for the assimilation of the indigenous population by Russians. At

(...continued) (October, 1975): 574-597; "Language Policy and Linguistic Russification"; and "The Ethnic and Language Dimensions."

39 Barbara A. Anderson, "Some Factors Related to Ethnic Reidentification in the Russian Republic," in Jeremy R. Azrael, Ed., Soviet Nationality Policies and Practices (New York: Praeger, 1978): 309-333; Anderson and Silver, "Estimating Russification."

18 the same time, the more extensive bilingualism (knowledge of Russian as a second language) among urban

Transcaucasians demonstrates the importance of the higher levels of contact between Russians and the indigenous populations in the urban areas (see Table 3, above).

Of the three titular nationalities of the Transcaucasian states, the Armenians were at greatest risk of assimilation because of their dispersion throughout the Soviet Union. Within Transcaucasia, moreover,

Armenians were more likely to be assured of Armenian schools, mass media, and religious facilities when they lived in Armenia. Moreover, the lack of opportunities to obtain higher education in Armenian would have encouraged many Armenian graduates of secondary schools in Azerbaijan and Georgia to move to Armenia if they wished to obtain further schooling in Armenian. Of course, complaints about insufficient cultural facilities in were high on the list of grievances by those who first demanded the uniting of Nagorno-Karabakh with Armenia, even if there were in fact Armenian schools in Nagorno-Karabakh at the time.

As of 1979, 4% of the Armenians in Nagorno-Karabakh claimed Russian as their native language

(while only 15 persons out of more than 123,000 Armenians in Nagorno-Karabakh claimed Azerbaijani as their native language). But 22% of all Armenians in Azerbaijan claimed Russian as their native language, and

40% of Armenians in Baku claimed Russian as their native language. Of those Armenians residing in Baku who claimed Russian as their native language, less than one-third claimed command of Armenian as a second language. Among Armenians who lived in Georgia in 1979, 9% claimed Russian as their native language; among Armenians who lived in Tbilisi (roughly one-third of the Armenians in Georgia), 19% claimed Russian as their native language, and of those, less than one-third claimed command of Armenian as a second language. Thus, judging by linguistic evidence from the censuses, substantial numbers of the Armenians in

Baku and Tbilisi were russified – claimed Russian as a native language and did not speak Armenian well.40

Russified Armenians in Azerbaijan and Georgia might normally have had little reason to migrate to Armenia even to obtain higher education. In fact, russified Armenians could well seek educational and

40 Of course, this does not mean that they could not "reclaim" Armenian language. Moreover, we are speaking here only of linguistic russification, not a change in ethnic self-identification (even if it carries with it broader implications for cultural identity and association preferences).

19 career opportunities in Russia and elsewhere in the Soviet Union. But those who were not russified would be more likely to choose Armenia as a destination to advance the opportunities for their children in their native language or to enter higher education themselves.

That Azerbaijanis may have been leaving Armenia and Georgia for Azerbaijan during the period before 1988 could have a similar explanation – the desire to obtain education and to advance careers in the context of one's national culture and language – but the potential pull of Azerbaijanis to Azerbaijan was probably moderated considerably by the low levels of urbanization of Azerbaijanis in Armenia and Georgia, where only 11% and 18%, respectively, of the Azerbaijani populations lived in urban areas in 1959, and only

10% and 20%, respectively, were urban in 1979. In fact, in contrast to the Armenians living outside of

Armenia, who were primarily urban residents (except those in Nagorno-Karabakh), Azerbaijanis living in

Transcaucasia outside of Azerbaijan were primarily rural.

Population Change in Nagorno-Karabakh

Table 6 shows the ethnic composition of Nagorno-Karabakh at recent census dates. Although at all dates, the vast majority of Nagorno-Karabakh was comprised of Armenians, the share of Azerbaijanis in the population increased rapidly from 1959 through 1979 and then declined by 1989.

[Table 6 About Here]

From 1959 through 1979, about 25% of all Armenians living in Azerbaijan resided in Nagorno-

Karabakh; in 1989, 37% of all Armenians in Azerbaijan lived in Nagorno-Karabakh. Recent reports suggest that by 1993 Armenians in Azerbaijan were even more concentrated in Nagorno-Karabakh.41 This was caused both by departures of Azerbaijanis from Karabakh and by the arrival of Armenian refugees there from elsewhere in Azerbaijan. Even though Nagorno-Karabakh was a site of conflict, it was apparently viewed after

1988 as more hospitable to Armenians than other parts of Azerbaijan.

41 USCR, Faultlines of Nationality Conflict.

20 Population Distribution and Movements after 1989

For the period after the 1989 census, we have no systematic counts of the population of

Transcaucasia. There are no official counts of persons killed due to wars or civil violence in Nagorno-

Karabakh, elsewhere in Azerbaijan, in Armenia, or in Georgia (including Abkhazia and southern Ossetia).

But the number of persons killed is far smaller than the number of displaced persons and refugees. Reports of the number of refugees by place of origin and by nationality are incomplete. Sometimes the number is defined in terms of officially designated refugees rather than by a count of persons who may have been forced to move or who sought asylum.42 Sometimes the reported figures do not distinguish between the "stock" of refugees and the "flow" of refugees, so that it is difficult to know when and in what numbers the different refugee streams came. Sometimes refugees are registered in more than one place, so that there can be double- counting. And sometimes refugees to an area subsequently take refuge elsewhere, to seek work or material support or perhaps to return to their place of origin when conditions improve.43

Nonetheless, based on information reported to the United Nations High Commission for Refugees

(UNHCR), we have a rough picture of the main movements of refugees. By mid-1992, the Russian Federation had 45,500 registered Armenian refugees (presumably mostly from Azerbaijan), 8,000 registered Azerbaijani refugees (presumably mostly from Armenia), 113,000 Ossetians from Georgia, and 100,000 Russians from other countries of the former Soviet Union.44 Many of the Russians came from Transcaucasia, in particular from Azerbaijan and Georgia (including Abkhazia and South Ossetia).45

42 To give an idea of the range of uncertainty, which has partly to do with definitions and partly to do with inadequate sources, it has been reported that as of January 1, 1994, there were 450,000 registered refugees and forced migrants in Russia from the other former Soviet republics. But the director of the Federal Migration Service in Russia estimated in Fall of 1993 that there were 2 million forced migrants in Russia. See Galina S. Vitkovskaia, "Russians in the Non-Russian Former Republics and Forced Migration to Russia," Presented at the Radford Conference on the Geodemography of the Former Soviet Union, Radford University, Virginia (August, 1994).

43 USCR, Faultlines of Nationality Conflict.

44 United Nations, Executive Committee of the UN High Commission for Refugees (UNHCR), UNHCR Activities Financed by Voluntary Funds: Report for 1991-92 and Proposed Programmes and Budget for 1993 (New York: United Nations, 1992).

45 John B. Dunlop, "Will a Large-Scale Migration of Russians to the Russian Republic Take Place over the Current Decade?" International Migration Review 27 (Fall, 1993): 605-629. We do not have figures for the places of origin of (continued...)

21 At the same time, Armenian officials reported to the UNHCR that 270,000 Armenians had fled

Azerbaijan during the period 1988 to mid-1992.46 The number of Armenians reported to have fled Nagorno-

Karabakh was 42,000. A later report put the total number of Armenians who had left Azerbaijan for Armenia between 1988 and 1990 at 260,000, and the number who had left Nagorno-Karabakh and the Shaumian region in 1992 and 1993 at 80,000, as the Azerbaijani National Army achieved considerable success in the region;47 a still later report mentioned that nearly 40,000 Armenians had returned from Armenia to Nagorno-Karabakh or regions between Karabakh and Armenia in 1993 and the first quarter of 1994, after Armenian armed forces had reversed their military fortunes in the region, while many others had gone to Russia because of economic hardships in Armenia.48 By the beginning of 1994, all Azerbaijanis had reportedly departed Nagorno-

Karabakh and about 20,000 Armenians originally from the region no longer lived there.49

Azerbaijani authorities reported that 217,000 Azerbaijanis had fled to Azerbaijan from Armenia as of mid-1992. Azerbaijan also had received about 50,000 who fled Uzbekistan in 1989

(another 42,000 Meskhetians reportedly went to Russia). No figures were available on the number of

Azerbaijanis displaced due to the fighting in Nagorno-Karabakh.50 But as of December 31, 1993, the

Azerbaijani government reported that there were in all 778,000 internally displaced persons resulting from the conflict over Nagorno-Karabakh,51 roughly one out of every ten persons in Azerbaijan. While most of the

(...continued) the Russian refugees and displaced persons.

46 UNHCR 1992.

47 UNHCR, UNHCR Activities Financed by Voluntary Funds: Report for 1992-93 and Proposed Programmes and Budget for 1994 (New York: United Nations, 1993).

48 UNHCR, UNHCR Activities Financed by Voluntary Funds: Report for 1993-94 and Proposed Programmes and Budget for 1995 (New York: United Nations, 1994).

49 Stephan H. Astourian, "The Nagorno-Karabakh Conflict: Dimensions, Lessons, Prospects," Mediterranean Quarterly 5 (Fall 1994): 85-109.

50 UNHCR 1992. A later report (UNHCR 1994) put the total number ethnic Azerbaijani refugees plus Meskhetian Turks at 230,000 as of December 31, 1993. We do not have an explanation for the difference from the previously reported number.

51 UNHCR 1994.

22 refugees from Armenia by then had been permanently settled in Azerbaijan, four-fifths of those displaced as a result of the Nagorno-Karabakh conflict had not been settled by the end of 1993.

Based on available information, it appears that almost all Armenians had left Azerbaijan by early

1991 except those who still resided in Nagorno-Karabakh or in a couple of nearby districts (Shaumian and

Khanlar). Virtually all Azerbaijanis had been forced out of Armenia. In addition, perhaps several tens of thousands of Russians left Azerbaijan and the other Transcaucasian republics in the late 1980s and early

1990s. Thus, the patterns of migration flow in Transcaucasia between 1959 and the late 1980s were changed sharply in degree by the Armenian-Azerbaijani conflict that erupted in 1988, but not so much in direction.

However, the estimated 113,000 Ossetians who had been forced to migrate to Russia by mid-1992 did involve a change in direction; they constituted two-thirds of the Ossetians in Georgia.52

Conclusion

Tragic recent events have focussed attention on the population of Transcaucasia. Without the events in Nagorno-Karabakh and since, ethnic antipathies in that region might be ignored. However, we see evidence of Russians leaving this region since 1959, and members leaving those republics in which they were not dominant since about 1970. When the environment became uncomfortable for a member of a minority group, returning to a region in which one's group is dominant was not always the choice. These may have been members of mixed marriages and they or their spouse may not have even known their nationality's traditional language. One result of Soviet language policy was to induce a high level of Russian language knowledge among people who did not live in their nationality's traditional area. In an era of increased ethnic and linguistic awareness, Russia seemed more hospitable than the group's traditional homeland.

At the same time that Russia was increasingly the destination for migrants, or, more accurately speaking, refugees, Russia's economic and social problems and were growing.53 The very magnitude and

52 According to the 1989 census, 162,000 Ossetians resided in Georgia. Of those, 65,000 lived in the South Ossetian Autonomous Province, and 97,000 lived in Tbilisi or its environs (33,000 in Tbilisi proper). Thus, the number of Ossetians forced to leave Georgia by mid-1992 substantially exceeded the number who lived on the territory of the South Ossetian Autonomous Province.

53 Tat'iana Regent, "Migratsiia v Rossii: Real'nost' i politika" [Migration in Russia: Reality and Policy]. Presented at (continued...)

23 diversity of the refugee problem led to increasingly less welcoming and favorable treatment of refugees, and a host of obstacles to the resettlement and humanitarian treatment of refugees appeared, such as lack of choice in place of residence, restrictions on residence permits in large cities, restrictions on entrepreneurship, and the simple lack of employment opportunities in a generally unfavorable economic climate in the late 1980s and early 1990s.54 Refugees and recent migrants unable to find housing or stable and legal work had to seek other means of support, to move elsewhere, or to emigrate abroad. As Russian nationalist concerns become even greater, the fate of these refugees is not likely to improve. The choice, then, for many refugees could increasingly become one of selecting the least bad situation rather than the most desirable situation.

(...continued) the Conference on Current and Emergent Migration within and from the Former USSR: Domestic and International Policy Perspectives, The Hague (March 4-5, 1993).

54Zhanna Zaionchskovskaia, "Recent Internal Migration Flows in Russia," Presented at the Radford Conference on the Geodemography of the Former Soviet Union, Radford University, Virginia (August, 1994).

24 TABLE 1 Population Increase of Major Transcaucasian Nationalities in USSR and Transcaucasian Republics: 1959–1989 ______

Population Percentage Change (in thousands)

1959 to 1970 to 1979 to 1959 1970 1979 1989 1970 1979 1989 ______

In Entire USSR

Armenians 2,786.9 3,559.2 4,151.2 4,623.2 + 27.7 + 16.6 +11.4 Azerbaijanis 2,939.7 4,379.9 5,477.3 6,770.4 + 49.0 + 25.0 +23.6 Georgians 2,692.0 3,245.3 3,570.5 3,981.1 + 20.6 + 10.0 +11.5 Russians 114,113.6 129,015.1 137,397.1 145,155.5 + 13.1 + 6.5 +5.7 All 208,826.6 241,720.1 262,084.7 285,742.5 + 15.8 + 8.6 +9.0

Armenia

Armenians 1,551.6 2,208.3 2,725.0 3,083.6 + 42.3 + 23.4 +13.2 Azerbaijanis 107.7 148.2 160.8 84.9 + 37.6 + 8.6 -47.2 Georgians 0.8 1.4 1.3 1.4 + 76.3 -7.1 +7.7 Russians 56.5 66.1 70.3 51.6 + 17.0 + 5.9 -25.7

Azerbaijan

Armenians 442.1 483.5 475.5 390.5 + 9.4 - 1.8 -17.7 Azerbaijanis 2,494.4 3,776.8 4,708.8 5,805.0 + 51.4 + 24.7 +23.3 Georgians 9.5 13.6 11.4 14.2 + 43.2 -16.2 +24.6 Russians 501.3 510.1 475.3 392.3 + 1.8 - 6.9 -17.5

Georgia

Armenians 442.9 452.3 448.9 437.2 + 2.1 - 1.0 -2.5 Azerbaijanis 153.6 217.8 255.7 307.6 + 41.8 + 17.5 +20.3 Georgians 2,600.6 3,130.7 3,433.0 3,787.4 + 20.4 + 9.7 +10.3 Russians 407.9 396.7 371.6 341.2 - 2.7 - 6.2 -8.9 ______TABLE 2 Regional Distribution of Major Transcaucasian Nationalities: 1959–1989 ______

Percent of nationality located in:

Own Other Transcaucasian Republic Russia Rest of Republic USSR Armenia Azerbaijan Georgia ______

Armenians 1959 55.7 — 15.9 15.9 9.2 3.3 1970 62.0 — 13.6 12.7 8.4 3.3 1979 65.6 — 11.4 10.8 8.8 3.3 1989 66.7 — 8.4 9.5 11.5 3.9

Azerbaijanis 1959 84.9 3.7 — 5.2 2.4 3.8 1970 86.2 3.4 — 5.0 2.2 3.2 1979 86.0 2.9 — 4.7 2.8 3.6 1989 85.4 1.3 — 4.5 5.0 3.5

Georgians 1959 96.6 0.0 0.4 — 2.1 0.9 1970 96.5 0.0 0.4 — 2.1 1.0 1979 96.2 0.0 0.3 — 2.5 1.0 1989 95.2 0.0 0.4 — 3.3 1.2 ______TABLE 3 Population Size and Urbanization of Transcaucasian Republics: 1926–1989 ______

Armenia Azerbaijan Georgia USSR

Total Percent Total Percent Total Percent Total Percent Pop. Urban Pop. Urban Pop. Urban Pop. Urban (x 1,000) (x 1,000) (x 1,000) (x 1,000) ______

All Population

1926 881 19 2,314 28 2,677 22 127,028 18 1959 1,763 50 3,698 48 4,044 42 208,827 48 1970 2,492 59 5,117 50 4,686 48 241,720 56 1979 3,031 66 6,028 53 5,015 52 262,436 62 1989 3,305 67 7,021 54 5,401 55 285,743 66

Titular Nationality

1926 20 17 16 1959 52 36 35 1970 63 41 43 1979 69 46 48 1989 69 50 53

Russians

1926 25 80 73 1959 71 88 79 1970 79 92 83 1979 83 94 85 1989 85 95 86 ______TABLE 4 Indigenous Ethnic Dominance in Transcaucasian Republics: 1959–1989 ______

Titular Nationality as Percent of Population (In parentheses: Russians as percent of population) All Urban Rural ______

Armenia 1959 88 (3) 92 (5) 84 (2) 1970 89 (3) 93 (4) 82 (1) 1979 90 (2) 95 (3) 81 (1) 1989 94 (2) 96 (2) 88 (1)

Azerbaijan 1959 67 (14) 51 (25) 82 (3) 1970 74 (10) 61 (18) 87 (2) 1979 78 (8) 68 (14) 89 (1) 1989 83 (6) 77 (10) 90 (1)

Georgia 1959 64 (10) 53 (19) 73 (4) 1970 67 (8) 60 (15) 73 (3) 1979 69 (7) 64 (12) 73 (2) 1989 70 (6) 68 (11) 73 (2) ______TABLE 5 Knowledge of Russian by Urban and Rural Populations of Transcaucasian Nationalities: 1959–1989 (in percents) ______Armenia Azerbaijan Georgia Urban Rural Urban Rural Urban Rural ______Knowledge of Russian as Native Language Armenians 1959 1.3 0.1 23.1 0.9 13.0 0.9 1970 0.3 0.0 22.8 0.3 13.7 0.7 1979 0.7 0.2 29.3 0.9 15.3 1.0 1989 0.3 0.3 20.2 0.8 14.8 1.0 Azerbaijanis 1959 2.2 0.1 2.1 0.1 5.7 0.2 1970 1.3 0.1 1.8 0.0 6.2 0.1 1979 2.6 2.2 2.2 0.1 6.3 0.2 1989 1.4 0.0 0.8 0.1 4.6 0.2 Georgians 1959 24.5 8.0 45.1 0.1 1.1 0.1 1970 12.4 3.5 34.1 0.4 0.9 0.0 1979 22.5 7.5 43.7 0.3 0.9 0.1 1989 13.2 5.2 34.4 0.4 4.1 0.0

Knowledge of Russian as Second Language Armenians 1970 31.6 9.5 44.5 0.8 43.0 25.7 1979 41.7 17.7 48.2 26.2 44.6 37.9 1989 51.4 28.3 57.2 42.3 40.3 46.7 Azerbaijanis 1970 11.7 4.3 29.7 4.6 37.6 11.9 1979 17.2 9.0 41.7 16.1 47.9 20.9 1989 29.8 18.1 45.1 18.1 45.7 30.4 Georgians 1970 45.5 46.2 34.9 5.7 35.8 8.3 1979 39.5 44.4 34.7 16.5 38.7 13.4 1989 37.2 46.7 41.5 19.9 39.8 22.7 ______TABLE 6 Population Composition of Nagorno-Karabakh: 1959–1989

Number Percent of Region ______

Azerbaijanis 1959 17,995 13.8 1970 27,179 18.1 1979 37,264 22.3 1989 40,668 21.5

Armenians 1959 110,053 84.4 1970 121,068 80.5 1979 123,076 75.9 1989 145,450 76.9

Russians 1959 1,790 1.4 1970 1,310 1.0 1979 1,265 0.8 1989 1,922 1.0 ______