Sie sind auf Seite 1von 17

Intermarriage between Turks and Kurds in Contemporary Turkey: Inter-Ethnic Relations

in an Urbanizing Environment
Author(s): Ayşe Gündüz-Hoşgör and Jeroen Smits
Source: European Sociological Review, Vol. 18, No. 4 (Dec., 2002), pp. 417-432
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/3559545
Accessed: 12-06-2018 02:56 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/3559545?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access
to European Sociological Review

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
European Sociological Review, Vol. 18 No. 4, 417-432 417

Intermarriage between Turks and Kurds


in Contemporary Turkey
Inter-ethnic Relations in an Urbanizing Environment

Ayje Giindfii-HojgorandJeroen Smits

Data from the Turkish Demographic and Health Surveys are used to study Turkish-Kurdish inter-
marriage in contemporaryTurkey.We find a strong tendency to marry within the own ethnic group
(or towards 'ethnic homogamy') which, however, has decreased significantly between the early
1960s and the late 1990s. So, in spite of the conflict in south-eastern Turkey, the groups seem to
have grown together somewhat in recent decades. This result remains intact if we control for the
difference in group size between the groups and for the presence of educational homogamy. An
individual-level analysis shows that most Turkish-Kurdish intermarriage takes place between
Kurdish males and Turkish females and that bothTurks and Kurds intermarry more in the large cities
and in regions where their own group is small. With regard to education, the highest intermarriage
tendencies are found among Turks with a low educational level and among Kurds with a high educa-
tional level. This finding is in line with social exchange theory.

Introduction

The degree of intermarriage between the members factors by which this relationship is influenced and
of different groups in a society is an important indi- to monitor the changes over time. The study of inter-
cator of the strength of the social boundaries marriage is one of the ways in which this can be
between those groups. From the first half of the done.

twentieth century on, information on intermarriage In this paper, we use data from the 1993 and 1998
has been used to gain insight into the structure of Turkish Demographic and Health Surveys to
ethnic and racial relationships and into the social explore the patterns and trends in Turkish-Kurdish
cohesion of societies (Drachsler, 1921; Merton, intermarriage between the early 1960s and the late
1941; Gordon, 1964; Blau and Schwarz, 1984; 1990s. We test hypotheses on the effects of modern-
Alba and Golden, 1986; Kalmijn, 1998). Within ization and on the exchange of favourable and
this literature, however, few past studies focus on unfavourable characteristics on the marriage market.
intermarriage between the Turks and the Kurds inBecause in Turkey the ethnic division to a certain
Turkey. Moreover, the existing studies are mainly extent goes together with an educational division
anthropological case-studies (e.g. Andrews, 1989). (Kurds have on average a lower educational level
Up to now no empirical research has been carried than Turks), the tendency to marry within one's
out at the nation-wide level. Given the social own ethnic group (or towards ethnic homogamy)
dynamics which have plagued the relationship
might to a certain extent be a by-product of the
tendency
between Turks and Kurds throughout the history of to marry within one's own educational
group
the region, it is important to gain insight into (or towards educational homogamy). This
the

? Oxford University Press 2002

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
418 GUNDUZ-HO?GOR AND SMITS

by-product explanation (Ultee, Dessens, andJansen, Kurds and the ruling class of the Empire's forces
1988) is tested in a multivariate log-linear analysis. (Kiri?;i and Winrow, 1997). During World War I
In the following sections of the paper we present and the following War of Independence (1919-23)
theoretical background information on the situa- the conflicts between the Kurds and theTurkish sup-
tion of the Kurds in Turkey, on intermarriage and porters of the new (Kemalist) regime were put aside.
its use as an instrument for gaining insight into the They allied against the French, Italian, Greek, Rus-
relationships among (ethnic) groups, on primordial- sian, Armenian, and English forces and the Kurdish
ism, and on modernization. In the empirical part of tribal and religious leaders participated in the major
the paper, we first present descriptive information decision-making processes at the Turkish National
on the situation of the Turks and Kurds in Turkey Assemblies.

and on Turkish-Kurdish intermarriage. Then we When in 1923 the Ottoman Empire was dissolved
present the results of a multivariate logistic regres- in favour of a newAnatolian-based nation-state, the
sion analysis in which the effects of selected Turkish Republic, under the leadership of Mustafa
characteristics of the partners on their intermarriage Kemal Ataturk and his Republican People's Party
tendencies are studied. Finally, a multivariate log- (RPP), new conflicts arose (Berkes, 1964). The main
linear analysis is performed in which the trend in aim of the regime was to transform the country into a
ethnic intermarriage is determined, independent of secular 'modern' industrial nation-state. To realize
the effect of educational intermarriage and of the this reconstruction, major modifications were intro-
differences in group size between the ethnic groups. duced into the political, economical, legal, and ideo-
logical structures. The role of Islam was reduced
drastically and special attention was paid to the
The Kurds in Turkey
position and legal rights of women. With the help
After the Turks, the Kurds form the major ethnic of these reforms, the Kemalist regime tried to pull
group in Turkey. The large majority of Kurds are the country out of its peripheral status. The spread
Muslims, but there are major divisions with regard of rational and secular thinking was advocated,
to denomination. Moreover, there are three main this being seen as the basis of industrialization
language dialects in Kurdish: Kirmanc, Zaza and (Gunduz- Ho,gor, 1997).
Soran (Van Bruinessen, 1992; Giine?-Ayata and The modernization attempts were relatively
Ayata, 1998). By 1990 about 12 to 13 per cent of successful in accelerating economic development
the Turkish population belonged to the Kurdish- in the western regions. However, in the east, some
(dialects)-speaking ethnic groups (Mutlu, 1996). of the Kurdish tribal and religious leaders did not
For many centuries, the Kurds tended to live in welcome the changes and revolted against the new
the high mountains and plains of the south-eastern regime (Tokluoglu, 1995). After suppressing these
and eastern parts of Turkey, where the climate is resistance movements, in 1938 the Turkish govern-
rather unfavourable (very hot summers and very ment initiated a number of policies which had a
cold winters). Until recently, an important character- significant impact on the Kurds. First, only Turkish
istic of Kurdish society was its tribal structure and was recognized as an official language. This was an
many Kurds were living under the authority of attempt to unite the country, where 29 different
their religious leaders (heikhs) (Van Bruinessen, languages were spoken i.e. to create a new nation-
1992). state without ethnic divisions. Secondly, the Latin
The roots of Kurdish ethnic problems in Turkey script was initiated. Thirdly, by law compulsory and
go back to the mid-19th century (Yegen, 1999). free primary education for all children was insti-
When the Ottoman Empire came under the eco- tuted. However, until very recently this education
nomic domination of European powers many was only in Turkish and schooling in Kurdish was
provinces declared their independence from the not allowed. From the Kurdish point of view, these
Empire (Ahmed, 1993). In the east, some of the measures meant that their mother-tongue was not
Kurdish tribal and religious leaders initiated resis- officially recognized. Speaking Kurdish at home
tance movements, which later turned into acts of but training in Turkish at school might hamper
rebellion and led to a sort of civil war between the the Kurdish children's cognitive development and

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
INTERMARRIAGE BETWEEN TURKS AND KURDS IN CONTEMPORARY TURKEY 419

decrease their chances of moving into the higher is not merely a simple process of random meeting
ranks of society. or spontaneous falling in love, but also, and in
During different periods (until 1991) publishing many societies primarily, a rationally and carefully
and broadcasting in Kurdish were forbidden and guided transaction in which important economic
giving Kurdish names to children is still not allowed and social considerations play a role and many
(Icduygu et al., 1999). According to Icduygu et al. more persons than the partners and their close rela-
(1999) this political and cultural repression created tives may be involved.
an insecure environment for the Kurds. The largest Given the important social functions of marriage
problems, however, were economic. There were in many societies, social scientists have long used
limited investigations into the infrastructure of the intermarriage data to gain insight into the quality
of the relationships among ethnic, religious, or
south-eastern region and not enough economic
immigrant groups (e.g. Drachsler, 1921; Jansen,
measures were taken for the region's social develop-
ment. 1982; Alba and Golden, 1986; Lee, 1988; Shavit and
Until the 1950s, many Kurds continuedStier, 1994; Smits,
to work as 2000). The existence of many
marital
sharecroppers or agricultural workers under the ties between the members of groups indi-
cates that there
authority of the tribal leaders or aghas (big land- are also other social contacts - like
owners). From the 1950s onwards, the friendships
eastern and - between them. Such ties may indicate
south-eastern regions were affected by
thateconomic
the children of the groups have the opportunity
change. With the introduction of agricultural
to meet each other - at school, in the neighbour-
machinery, many sharecroppers turned hood,
intoorseaso-
during leisure activities. And, most
nal agricultural workers (Akcay, 1995).importantly,
Over time, if intermarriage among the members
the land was also broken up into many of tiny equal
different groups is a common phenomenon, this
shares, because of the great number of indicates
siblings.that In
the members of these groups accept
most cases these lands were too small to each other as an
support social equals (Kalmijn, 1998).
extended family. Therefore, many Kurds Once
begantheretoare mixed marriages among the
members
migrate to the main urban centres of western of different groups in a society, another
Turkey
or, from the 1960s onwards, toWesternaspect of intermarriage
Europe (SIS, begins to play a role: its
function
1994; Ak?it etal., 1996). In later years also, as a connecting element in a society.
conflict
between the Turkish security forces and Mixed
themarriages
separa- not only link together two indi-
viduals, but also the larger groups to which these
tist Kurdish Workers Party (PKK) in southeastern
individuals
Turkey led to the migration of many people belong. Such marriages form a bridge
to cities
between
in the region and to the West. These changes these larger groups over which family
worked
as a powerful instrument in breaking upmembers and friends of the partners may come
the primor-
dial tribal structures. On the other hand, this into contact with each other and new - group-
confrontation with outsiders for the many Kurdsboundary-transcending - personal contacts and
who had never been more than a few neighbouringcollaborations may come into existence. For this
villages from home, may also have strengthenedreason, intermarriage is expected to promote the
their Kurdish identity and nationalistic feelings. social cohesion of societies in which different ethnic
groups live together, or which consist of a number of
smaller units, like clans (e.g. Levi-Strauss, 1969; Van
den Berghe, 1979; Blau and Schwartz, 1984; Smits,
Intermarriage
2000).
Marriage can be defined as a socially recognized
'contract' to form a family. In almost every society
Factors Influencing Intermarriage
marriage is considered to be a very important social
event, and marriage relationships are in general To gain insight into the factors that affect the
strongly regulated and protected by social controlstrength of the tendency towards intermarriage
mechanisms, family laws, and/or religious rules.among ethnic groups, it is important to study the
marital decisions of the individuals belonging to
This illustrates that 'who is going to marry whom',

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
420 GUNDUZ-HO?GOR AND SMITS

these groups. After all, the boundaries between homogamy in a region where everybody belongs to
ethnic groups, which are observed at the aggregate the same ethnic group. In such a region ethnicity
level, are the outcome of the marriage choices of probably plays no role at all in partner choice,
many individual group members. In western because it is not a criterion on which groups can be
societies, the preferences of the partners are usually identified. A more interesting point is to understand
central in the process of partner choice, but in the how it is possible that there is ethnic homogamy
Middle-Eastern Muslim countries other social within large multi-ethnic cities, like Istanbul.
Given the fact that many Kurds migrated from
group members, especially parents and relatives
(who play the role of 'matchmaker') are much more
their traditional living places in the east of Turkey
involved in marriage decisions. to the large urban areas, this is a very relevant issue.
Wetheir
Besides the preferences of the partners and would like to know how it is possible for these
families, the outcomes of marital decisions Kurds
are also
keep their ethnic identity there, at least to a
influenced by structural characteristics certain
of the extent. A possible answer to this question
can be
marriage market (Kalmijn, 1998). An important found in the literature on the primordial
fac-
basisbeing
tor in this respect isgroupsite. All other things of ethnic identity.
equal, the probability of a coincidental meeting with
somebody from a large group is higher than the
The Primordial Basis of Ethnic Identity
probability of such a meeting with somebody from
a small group (e.g. Blau and Schwartz, 1984). The term primordial may be defined as 'first devel-
Furthermore, the size of the smallest groups oped is
or acreated', but it can also be used to mean
'primeval' suggesting the existence of something
restrictive factor. The number of mixed marriages
cannot be higher than the number of persons from inthe beginning. The 'primeval' notion carries
with
the smallest group. If there are differences in groupit a strong suggestion that primordial attach-
size, there will always be a number of personsments
in the are rooted in the 'natural order of things'.
The naturalness or sacredness of primordialism is
largest group for which no partner in the smallest
connected to social ties that include blood, race,
group is available. For this reason, the proportions
of mixed marriages between the members language,of two customs, and religion (Calhoun, 1993;
Bobrow, 1996). Cultural elements, like language
groups do not give a good indication of the prefer-
ences for (or aversion against) a partner from and religion,
the constitute essential elements of the
cultural heritage of the ethnic group and assist in
other group. To gain insight into these preferences
for groups that differ greatly in group size, like the the ethnic identity to the next genera-
transmitting
tions.
Turks and the Kurds in Turkey, we have to use mea-At the same time, ethnic identity is a socially
constructed
sures that are insensitive to differences in group size, definition of the self, whose meaning
may change in the course of time. New ethnic
like odds ratios or log-linear parameters (Kalmijn,
1998). groups are continuously being created which receive
Another important structural factor is spatialsegre- primordial status. So ethnic identities are not static
gation.We cannot expect many marriages between the constructs. They are 'renewed, modified and remade
members of groups which are concentrated in differ- in each generation' (Cornell and Hartmann, 1998).
ent parts of a country. And even if the geographical Ethnic group members tend to behave according
distances are not so large, there may be spatial segre- to their 'primordial' bases of affiliation in daily life.
gation, for example if the Turks tend to live in the This usually leads them to choose a marriage partner
cities, whereas the majority of the Kurds live in the from their own ethnic community.
countryside. The boundaries of ethnic groups can The primordialism literature also suggests that
persist even if the members of one group live side ethnic feelings may arise out of specific social condi-
by side with other groups and share the same geogra- tions. Being a member of an ethnic group, for
phical, political, economic, and social environment. example, may mean social and psychological soli-
This last issue brings us to the concept of cul- darity for a recent migrant in an urban centre. In
tural boundaries between ethnic groups. It is not addition, Bobrow (1996) argues that when an ethnic
very difficult to understand the existence of ethnic group is threatened or experiences opposition,

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
INTERMARRIAGE BETWEEN TURKS AND KURDS IN CONTEMPORARY TURKEY 421

ethnic feelings are generated or regenerated. Intense expected to increase the number of contacts between
suffering experienced in the past by the ancestors of the members of different social groups, to erode the
the people also have an important impact on the cultural barriers between them, and to reduce par-
formation and the continuation of the ethnic ents' ability to control the marriage choices of their
identities. The establishment and sustenance of children (Smits, Ultee, and Lammers, 1998). In this
primordial attachments to the community maytransformation
be process, modernization theory sug-
closely related to gender and language issues.
gests that ascribed status, which is 'assigned' to an
Women probably play a key role in passing individual
the by society at birth (such as gender, age,
primordial attachments on to future generations.
racial, and ethnic backgrounds) becomes less impor-
tant for the distribution of individuals over social
In closed community structures, Kurdish women
may experience the control of the patriarchal positions and achieved status, which is'attained' by
ideologies more strongly than men and, because of
a person's own efforts (such as through the means of
education and occupation) becomes more impor-
this, their access to the 'outside' world probably
tant (Blau and Duncan, 1967). Accordingly, in
remains rather limited. The patriarchal restrictions
may even hinder women's access to legitimate
studies on intermarriage, ethnicity (which is consid-
resources in the public sphere, such as education ered as an ascribed characteristic) is expected to
and employment opportunities, and keep them in
become less important in marriage choice whereas
education (which is considered as an achieved char-
the private domain. This may also restrict the contact
between Kurdish females and Turkish males, thus acteristic) becomes more important (e.g. Kalmijn,
reducing the likelihood of intermarriage by these 1991; Shavit and Stier, 1994).
Kurdish women. This more or less 'classical' theory of modern-
This isolation may be strengthened by the fact ization leads us to the expectation that, at the
that quite a large number of Kurdish women have aggregate level, the degree of ethnic homogamy in
Turkey would have decreased in the period under
no schooling at all and are not even able to speak
the dominant Turkish language (Smits and study, because of the ongoing modernization
Giinduz-Hosgor, forthcoming). These women, process of Turkish society. At the individual level,
we would expect that individuals who are more
therefore, may be much more dependent on primor-
dial values and traditions than their male'modern' (have a higher educational level, work in
counterparts, and consequently may playnon-agricultural
a major occupations, live in urban areas,
are of
role in passing these primordial attachments younger/born
the later) have a higher tendency
community on to future generations. towards intermarriage.
Given the central role of ethnicity in many of
the onflicts which are currently taking place in
different parts of the world, this 'classical'
Hypotheses modernization hypothesis does not seem to be
Modernization very realistic, or at least explains only part of the
story. In this paper we will therefore also test an
In the stratification literature, modernization is
alternative hypothesis on the effect of moderniza-
tion,
defined as a 'phased, irreversible, progressive, which predicts that modernization may lead
and
lengthy' process in which a society moves to an increase of ethnic consciousness which coun-
from
terbalances
traditional (or less developed) institutions to those the effects of the other aspects of
characteristic of more developed societies (Bell,
modernization in the area of partner choice. This
1981; So, 1990: 33). Modern societies tend to be is derived from the literature on the pri-
hypothesis
mordial basis of ethnic identity, discussed in the
urbanized, literate, and industrialized. Accordingly,
modernization assumes a 'process that moves insection, which states that ethnic identity
former
the direction of the western development maymodel'
be a constructed identity used by groups to
(So, 1990: 34). Various aspects of modernization -
strengthen their social position, and that latent eth-
like urbanization, increased geographic mobility
nic identities may become manifest in situations of
and more pervasive mass communication - are
uncertainty or threat.

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
422 GUNDUZ-HO?GOR AND SMITS

Earlier in this paper, we saw that the modern- difference in educational level between the Turks
ization of the countryside in the eastern part of and the Kurds. In this way, ethnic homogamy in
Turkey, in the second part of the twentieth century, Turkey might be strengthened by educational
went hand in hand with a strong migration stream of homogamy.
Kurdish peasants with very little education and The social exchange theory in the area of partner
knowledge of city life, to the large urban areas of selection (e.g. Davis, 1941; Merton, 1941; Kalmijn,
Western Turkey or Western Europe. The feelings of 1998; Schoen and Wooldredge, 1989), on the other
uncertainty of this group were most likely to have hand, focuses on the fact that the characteristics of
strengthened their need for identity and probably individuals may differ in their value in the marriage
made them more aware of the fact that they were market and on the possibility that partner selection
Kurds than they had ever been in their home villages may take place on the basis of more than one char-
(in those home villages the persons they met were acteristic. This theory states that in the marriage
'normal others', but meeting the same people in market people may compensate unfavourable char-
Istanbul would be an encounter with 'fellow acteristics in one respect with more favourable
Kurds'). Depending on the strengthcharacteristics
of this in other respects. With regard to eth-
nic intermarriage this means that individuals whose
'opposite' force, the increase in ethnic intermarriage
predicted by the 'classical' modernization social
theory
status in society is low because they belong to
may have been slowed down to a greater an orethnic
lesserminority group, have better chances of
extent or even completely neutralized. marrying a member of the higher-prestige dominant
group if they can offer the partner a high socio-eco-
nomic status. In the same way, individuals with low
By-product and Exchange Hypotheses
socio-economic status belonging to the dominant
Besides hypotheses on the effects of modernization,
group may marry a partner from a higher socio-eco-
there are two other hypotheses which will nomic stratum from a minority group. From this
be tested
in this paper: the by-product hypothesiskind and a
of exchanges both partners may profit.
Translated
hypothesis on the exchange of favourable and un- to theTurkish situation, with theTurks
favourable characteristics in the marriagebeing the dominant ethnic group with the highest
market.
ethnic status, this means that we expect to find
Both hypotheses follow from the understanding
more
that people select their marriage partner not onethnic
the intermarriage among Kurds with
basis of only one characteristic, but on thehigher
basissocio-economic
of status and among Turks
with lower socio-economic status. The theory also
many characteristics, and that these characteristics
may be related to each other (having a high levelthe
predicts ofexistence of a gender asymmetry in the
education and a high level of income) or may differ
pattern of marital exchange, because traditional sex-
(having a high level of education and a lowroles make
level ofit difficult for males to marry females
income). with higher socio-economic status. This might
The by-product hypothesis takes the possible reduce the number of marriages between higher
relatedness of the characteristics of individuals as educated Kurdish females and lower educated
Turkish
its point of departure. Because of this relatedness, a males.

resemblance between spouses with regard to one


characteristic may be the result of partner selection
Data
on the basis of another characteristic. For example, and Method
Ultee etal. (1988) hypothesized that homogamy with
Data
regard to the (un)employment of spouses may be a
by-product of partner selection on the basisToof analyse the patterns of ethnic and educationa
education and the relationship at the individual
intermarriage, we use data from the 1993 and 19
level between education and (un)employment. In
Turkish Demographic and Health Surveys (TDHS)
These surveys are part of the Demographic an
the same way, ethnic homogamy among the Turks
Health Surveys (DHS+) programme which provid
and Kurds in Turkey may be a by-product of partner
selection on the basis of education and the large
data and analyses on the population, health, and

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
INTERMARRIAGE BETWEEN TURKS AND KURDS IN CONTEMPORARY TURKEY 423

nutrition of women and children in developing years. Forgender a dummy variable is used coded '1'
countries. The DHS+ surveys use nationally repre-for males and '2' for females. For the educationallevel of
sentative samples of households and consist of athusbands and wives, we recoded the educational
minimum a household survey and a women's survey variables used in theTDHS into the following four
(DHS+, 2002). In the current paper, we use the categories:
background characteristics of the women, their 1. No education or incomplete primary education;
2. Completed primary education;
husbands, and their marriage which were gathered
3. Incomplete or completed secondary-level
in the women's survey. Because the number of
mixed marriages between Turks and Kurds is low, education; and
the data from the 1993 and 1998 TDHS were com- 4. More than secondary level education.
bined. For the multivariate log-linear analysis of ethnic and
Besides the women who were married at the time educational homogamy the educational variables are
of the survey, we also included divorced and reduced further to only two categories:
widowed women in our analysis, because for these 1. Completed primary or less; and
women all the necessary information about the2. More than primary.
former husband was available in the data-sets. The To indicate changes over time, we use three cohorts on
total number of women (and husbands) availablethe basis of the date of marriage:
1. Married before 1976
for our analyses is 12,671 (6,519 from the 1993 survey
2.
and 6,152 from the 1998 survey). Because we Married 1976-1987
excluded individuals with other than Turkish or 3. Married after 1987.
Kurdish ethnicity and individuals for whom infor-
To indicate the occupationalstatus of the couple at the
mation on ethnicity or educational level was time of the interview, in most cases the husband's
missing, there were 11,570 couples or 23,140 indi-
occupation is used. If the occupation of the husband
viduals remaining for the analyses. is unknown, the occupation of the wife is used
instead. Occupational status is coded into six
Variables categories:
1. Upper non-manual, including professional,
technical,
To identify the ethnic group of the respondents and and managerial occupations;
their partners, we follow Mutlu (1996) and 2.Iqduygu
Lower non-manual, including clerical and sales
et al. (1999), who used mother-tongue foroccupations;
this
3. Service
purpose.We use the questions in the women's surveyoccupations;
on mother-tongue of the respondent, on 4. Manual worker;
mother-
5. Farmers
tongue of her husband, and on mother-tongue ofand farm-workers; and
their parents. Individuals with Kurdish or6. Zaza as
Occupational status unknown.
their mother-tongue and individuals both of has four categories:
Urbanization
whose parents have Kurdish or Zaza as1.their
Capital, large city;
2. Small city;
mother-tongue are defined as Kurds. Individuals
3. Town;
with Turkish as their mother-tongue and with no and
more than one of their parents having Kurdish 4. Countryside.
or
Zaza as their mother-tongue are defined asFor Turks.
region four categories are also used:
1. West;
Individuals with mother-tongues other thanTurkish
or Kurdish/Zaza are left out of the analysis. By South;
2. defin-
ing all individuals both whose parents have 3.Kurdish
Central and North; and
4. East.
or Zaza as mother tongue as Kurds, we take care that
The but
respondents who are from Kurdish background regions Central and North are combined
because
who speak Turkish as mother tongue are not there are almost no Kurds living in th
wrongly defined as Turks. north of Turkey. Both urbanization and region
Besides ethnicity, a number of other variables
referareto the current place of residence of the couple
used. The age of the respondents is measured in
Ethnically mixedfamily background is measured with

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
424 GUNDUZ-HOSGOR AND SMITS

dummy variable with the value of '1' if one of the estimated in which these effects and their inter-
parents has Turkish as her/his mother-tongue and actions are studied together (compare Kalmijn,
the other Kurdish, and a value of '0' otherwise. 1991). The advantage of a log-linear analysis over a
Survey is a dummy variable coded '1' for the 1993 logistic regression analysis is that the associations
TDHS and '2' for the 1998 TDHS. between the characteristics of spouses can be
studied while controlling for differences in group
size. For this analyses, a 2*2*2*2*3*2 table (ethnicity
husband*
Method
ethnicity wife*education husband*education wife*
In the empirical part of this paper, we first present cohort*survey) is constructed. An indicator of
descriptive information about the distribution of survey is present both in the logistic regression
Turks and Kurds over the explanatory variables analyses and in the log-linear analysis to control for
used in the study. Next, we present information possible differences between the 1993 and 1998
about the association between the ethnic groups of surveys.
husbands and wives, together and separately for the
three marriage cohorts. In this descriptive phase, we
focus on percentages of mixed marriages and odds Results
ratios. The percentages of mixed marriages give an
Descriptive Analysis
idea of how many marital relationships between
Turks and Kurds were found amongst the respon- Table 1 presents the basic characteristics of the mem-
dents to the survey. These percentages are important bers of theTurkish and Kurdish couples selected for
from a social cohesion perspective, because they our analyses.We see that most of theTurks live in the
show the extent of cohesion-increasing marital ties West and Central/Northern regions of Turkey,
among the groups. However, percentages say less whereas the majority of Kurds live in the East. With
about the preferences of the members of both ethnic regard to urbanization, the differences are much
groups for (or the aversion against) a marriage with smaller: of both groups, about 14 per cent live in
somebody from the other group. The reason is that towns and roughly one-third in the countryside.
percentages are rather sensitive to differences in The Turks are somewhat more concentrated in the
group size among ethnic groups, which, in theTurk- capital and largest cities whereas the Kurds are
ish case, are very large. Therefore, we also present more concentrated in the smaller cities.
odds ratios, which show the strength of the ten- With regard to education level, there are large
dency to marry within one's own ethnic group inde- differences between Turks and Kurds. About one-
pendent of the differences in group size between the quarter of the Kurdish males and more than 70 per
groups (e.g. Kalmijn, 1998: 405). cent of the Kurdish females have not completed
To gain insight into the effects of the character- primary education. For the Turks these percentages
istics of the partners on the strength of their are 7 and 22 per cent, respectively. Only 2.8 per cent
intermarriage tendencies, we present the propor- of the Kurdish males and 0.5 per cent of the Kurdish
tions of mixed marriages broken down by these females have more than secondary education,
characteristics. We also study the effects of these against more than 10 per cent of the Turkish males
characteristics simultaneously, in a multivariate and almost 5 per cent of the Turkish females. The
logistic regression analysis. The dependent variable occupational status differences between Turks and
in this analysis is a dummy variable which takes the Kurds are smaller, but at the highest occupational
value of '1' if the respondent is married to somebody level the Turks are clearly over-represented. The
from the other ethnic group and a value of '0' other- very small proportion of Turks and Kurds from
wise. mixed family background may be an artefact of the
In the final analysis of this paper, we try to find use of mother-tongue as the indicator of ethnicity. In
out to what extent ethnic homogamy is a by-product many ethnically mixed families, probably only one
of educational homogamy, or vice versa. For this of the languages was spoken at home, so the children
purpose, a multivariate log-linear model is might not know or realize that the mother-tongue of

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
INTERMARRIAGE BETWEEN TURKS AND KURDS IN CONTEMPORARY TURKEY 425

one of their parents was different. However, the Table 1. Percentage distribution of married Turks and Kurds over

respondents who declared themselves to be of central variables in the 1990s (rowpercentages)

mixed family background probably were. Therefore,


Turks Kurds
this variable can still be used to test for the presence . . . . . . I. . . . . . . . . . . . . . . . . . . . .

of a mixed-family background effect. Region


Table 1 also shows that the tendency towards inter- West 40.9 15.1

marriage among theTurks is very low: only 2 per cent South 15.0 11.4

of them are married to somebody from another Central/North 36.7 10.0

ethnic group, and 1.4 per cent have a Kurdish East 7.4 63.5

partner. Among the Kurds, the tendency towards Total (N) 100 (19929) 100 (3210)
Urbanization
intermarriage is somewhat higher: almost 10 per
Capital, large city 27.3 11.6
cent of the married Kurds have a non-Kurdish
Small city 25.2 35.4
partner, and 8.4 per cent of these partners Town
are 14.5 13.5
Turkish. The differences in intermarriage percen- 33.0 39.5
Countryside
tages between Turks and Kurds are not surprising,
Total (N) 100 (19928) 100 (3211)
given the fact that there are many more TurksEducation
than males
Kurds in Turkey. Because intermarriage between
Less than primary 6.7 26.5
Turks or Kurds and members of the other (smaller)
Completed primary 50.9 50.2

ethnic groups inTurkey is very rare, it probably does


At least some secondary 31.6 20.5
not make much difference that we exclude these Tertiary 10.8 2.8

marriages from our analysis. Total (N) 100 (9948) 100 (1623)
Education females

Less than primary 22.1 71.8


Ethnic Homogamy Completed primary 55.0 24.7

At least some secondary 18.2 3.0


Table 2 presents information on ethnic homogamy. 4.8 0.5
Tertiary
We see that 2.4 percent, or one in 40 marriages, was
Total (N) 100 (9981) 100 (1588)
mixed according to the ethnic group of the partners.
Occupational status family
In the period under study there is an increase ofUpper non-manual 16.1 9.3

ethnic intermarriage, from 1.8 per cent among the Lower non-manual 10.6 8.5

couples married before 1976 to 3.0 per cent among Services 12.4 12.9

the couples married after 1988.The pattern of changeManual 39.6 42.5


Farm 16.5 19.8
is rather interesting. In the first cohort, most mar-
Not working/unknown 4.8 7.1
riages seem to have taken place between Kurdish
males and Turkish females. However, the increase Total (N) 100 (19929) 100 (3211)
Mixed family background
of the intermarriage tendency among Turkish males
No 99.7 99.4
and Kurdish females was much higher than the
Yes 0.3 0.6
increase among Turkish females and Kurdish Total (N) 100 (19929) 100 (3211)
males. As a result, in the third cohort the difference
Married to:
in frequency between the two marriage combina-
Turk 98.0 8.4
tions has almost disappeared. Kurd 1.4 90.5

The changes in the intermarriage percentages of Other 0.7 1.1

Total (N)
the different categories may be due to changes in the 100 (20124) 100 (3258)
strength of the group boundaries. However, because
percentages are sensitive to changes in the marginals
of the table, these changes may also be duetogether,
to the odds ratio has a value of 793. This
changes in the number of married male and female means that, on average, for both Turks and Kurds
the odds of marrying a partner from the own
Turks and Kurds in this period. To clarify the situa-
tion Table 2 also presents odds ratios, which are group is almost 800 times as high as the odds of
marrying a partner from the other group. The
insensitive to such group-size effects. For all couples

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
426 GUNDOZ-HOGOR AND SMITS

Table 2. Percentages ofethnically mixed marriages of Turkish and Kurdish males andfemales and odds ratios ofethnichomogamy

Marriage cohort
........................I.................................I...............

Before 1976 1976-1988 1989-1998 Total

Turkish males 0.7 1.2 1.7 1.2


Turkish females 1.4 1.5 1.8 1.5
All Turks 1.0 1.3 1.7 1.4
Kurdish males 9.4 8.9 10.2 9.4

Kurdish females 5.2 7.2 9.5 7.5


All Kurds 7.3 8.1 9.8 8.5
Total 1.8 2.3 3.0 2.4
Odds ratio 1354* 864* 520* 793*
N 6,116 9,984 7,038 23,138

*p < 0.001.

tendency to marry within one's own group was thus Kurds those with the highest educational level
very strong. From the cohort of couples married show the highest intermarriage rates. For both
before 1976 to the cohort of couples married after groups this effect is quite strong. The Turks with
1988, the value of the odds ratio decreased from less than primary-level education tend to marry a
1354 to 520. So, it seems likely that the boundaries Kurdish partner more than four times as often as
between the ethnic groups really have become theTurks with more than secondary-level education.
somewhat more open in recent decades. And the Kurds with more than secondary-level edu-
In Table 3 the percentages of mixed marriages for cation are more than four times as likely to marry a
Turks and Kurds are broken down by a number of Turkish partner as the Kurds with less than primary-
relevant variables. With regard to region, the pattern level education.

is rather interesting. Turks living in the East of The effect of the occupational status of the couple
Turkey are much more likely to marry Kurds than is not very strong. Among theTurks intermarriage is
Turks living in the other parts of Turkey. For the slightly higher for service workers, whereas among
Kurds the percentage of mixed marriages is lowest the Kurds upper non-manuals tend to intermarry
in the East and quite high in all three other regions. most and farmers least. For both ethnic groups,
So it seems that the tendency towards intermarriage growing up in a mixed family increases the tendency
is stronger among individuals who live in a region towards intermarriage strongly. Almost one-third of
where their own group is relatively small and the theTurks from a mixed family background and more
other group is relatively large. than 40 per cent of the Kurds from a mixed family
With regard to urbanization, the differences in background are married to somebody from the
intermarriage rates for the Turks are rather small, other ethnic group.
except that Turks living in the countryside tend to
intermarry somewhat less. For the Kurds the differ-
Logistic Regression Analysis
ences according to urbanization are larger. In the
capital and largest cities, almost a quarter of the Table 4 presents the results of logistic regression
married Kurds have a Turkish partner, whereas in analyses with the odds on ethnic intermarriage as
the smaller cities and towns this is around 10 per the dependent variable. Because the effects found
cent and in the countryside less than 5 per cent. in the bivariate analyses sometimes differed quite
The relationship between intermarriage and considerably between the ethnic groups, the models
educational level is also very interesting. Among are estimated separately forTurks and Kurds.
the Turks the lowest educated tend to be most likely Neither for Turks nor for Kurds are the effects of
to marry a Kurdish partner, whereas among the gender and age significant. The coefficient for

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
INTERMARRIAGE BETWEEN TURKS AND KURDS IN CONTEMPORARY TURKEY 427

Table 3. Percenta g esixedmarriagesamoni gTurksandKurdsby in the multivariate analysis. However, the high level
characteristics of the respondents (N=23140) of intermarriage among Kurds in the West is no
longer present. This may be caused by the fact that
Turks
r many Kurds who live in the West of Turkey are living
Region in Istanbul or other large cities, and that the high
West 1.0 16.3 level of intermarriage among the Kurds living
South 2.1 17.5 there, in the multivariate analysis, is caught by the
Central/North 0.7 16.2 urbanization variable. Indeed we see that in the
East 5.3
3.8 Capital and the larger cities the tendency toward
Total 1.4
8.5 intermarriage among Kurds is significantly higher
Urbanization
than average, whereas in the countryside inter-
Capital large city 1.5
22.0 marriage among Kurds is significantly lower than
Small city 1.8
Town 1.5 1 average. This time we see the same pattern for the
Countryside 0.9 5 Turks. They also tend to intermarry significantly
Total 1.4 8.5 more than average in the Capital and larger cities
Education and significantly less than average in the country-
Less than primary 2.3 5.3 side.
Completed primary 1.3 10.2 The linear cohort effect is significantly pos
At least some secondary 1.2 14.2 for the Turks. So the tendency among Tu
0.5
More than secondary 22.2 marry a Kurdish partner has increased in re
Total 1.4
8.5 decades. The cohort effect is not significant f
Occupational status couple Kurds, although the sign of the coefficient is a
Upper non-manual 1.3
13.7 positive. This may be caused by a lack of stat
Lower non-manual 1.2
Services 1.8 . power, due to the lower number of Kurds in
Manual 1.5 8.6 data-set, or by the presence of other variables
Farm 1.1 5.5 analysis the effects of which may have change
Unknown 1.1 4.4 time. The survey effect is significantly positiv
Total 1.4 8.5 the Turks, which indicates that in the 1998 su
Mixed family background Turks tended to intermarry more with Kurds t
No 1.3 8.3 in the 1993 survey.
Yes 32.3 42.1 To find out whether the effects of the expla
Total 1.4
8.5 variables differed between males and females
tested for significant interaction effects of gen
with all other explanatory variables. Only one
education1 is significant for both e thnic groups: a these interaction effects turned out to be sig
higher educational level leads to si ignificantly less the effect of mixed family background on
intermarriage among the Turks and to significantly intermarriage was significantly stronger for t
more intermarriage among the Kurc Is. This interest- ish females than for theTurkish males.We also
ing finding is in line with the result of the bivariate whether the effects of the explanatory va
analysis in Table 3. The occupation lal status of the changed over time, by testing for significan
couple has no effect on intermarriag ge, but having a tions with the linear cohort effect. However
mixed family background does. E 3oth Turks and these interaction effects was significant.
Kurds from mixed family backgrot inds tend to be
married significantly more often with a partner
Log-linear Analyses
from the other group.
With regard to region, the resu lit presented in Because of the possibility that ethnic homogamy is a
Table 3- that intermarriage amon g Turks is rela- by-product of educational homogamy, or vice versa,
tively high in the East whereas s intermarriage we conducted a multivariate log-linear analysis in
among Kurds is rather low there - - remains intact which ethnic and educational homogamy are

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
428 GUNDUZ-HOSGOR AND SMITS

Table 4. Logistic regression modelsfor the effectofselected independent variables on the odds of Turkish-Kurdish intermarriage

Turks Kurds

Variable B S.E. Exp(B) B S.E. Exp(B)


. . . . . . . .

Constant -4.41** 0.52 0.01 - 3.53** 0.55 0.03


Gender -0.01 0.13 0.99 0.08 0.15 1.08

Age 0.01 0.01 1.01 0.02 0.01 1.02


Educational level -0.54** 0.10 0.58 0.49** 0.10 1.63

Occupational status couplea


Upper non-manual 0.05 0.17 1.05 0.29 0.19 1.33
Lower non-manual -0.10 0.19 0.90 -0.13 0.20 0.88
Services 0.16 0.15 1.18 0.13 0.16 1.14
Manual 0.02 0.12 1.02 -0.00 0.12 1.00
Farm -0.00 0.20 1.00 0.34 0.20 1.41
Unknown -0.13 0.29 0.88 - 0.62* 0.30 0.54

Mixed family background 2.90** 0.30 18.1 2.00** 0.53 7.38


Linear cohort effect 0.36** 0.12 1.44 0.23 0.13 1.26

Regiona
West - 0.80** 0.12 0.45 0.02 0.13 1.02
South 0.25* 0.11 1.29 0.47** 0.12 1.59
Central/North - 0.77** 0.12 0.46 0.59** 0.14 1.81
East 1.32** 0.12 3.74 - 1.08** 0.12 0.34
Urbanizationa

Capital large city 0.65** 0.13 1.92 0.49** 0.14 1.63

Small city 0.06 0.11 1.06 -0.07 0.11 0.94


Town 0.01 0.13 1.01 0.27 0.14 1.31

Countryside - 0.73** 0.14 0.48 -0.69** 0.15 0.50


Survey 0.34* 0.16 1.40 -0.31 0.18 0.73
N 19917 3223

-2 Log Likelihood 2564.3 1617.4


Model chi-square/DF 318.4/17 249.3/17

aFor these variables deviation contrast coding is used.


*p < 0.05, **p < 0.01.

studied simultaneously. Given the possibility of marriage becomes a completely random process.
disturbing survey effects, a variable indicating the All other models suppose a tendency towards
survey (1993 or 1998) was also included in the log- homogamy. Model 2 and Model 3 test for ethnic
linear model. Table 5 presents a number of log-linear and educational homogamy, respectively, and the
models together with their fit measures. Given the fourth model tests for the presence of both forms
rather large number of 11,567 couples in the table, of homogamy. Model 5 tests whether ethnic homo-
we rely on the BIC statistic (Raftery, 1995) to select gamy differs between the cohorts and Model 6 does
the best model for description of the data. In gen- the same for educational homogamy. Models 7 to 9
eral, the model with the lowest BIC value is the one test whether the differences between the cohorts can
that should be selected. be modelled as a linear trends, and Models 10 to 12
Model 1 is the model of independence, whichtest for the presence of survey effects.
supposes that the frequency of occurrence of each The best-fitting model according to the BIC
marriage combination is completely determined bycriterion is Model 10, which assumes the presence
the numbers of men and women in the margins of of both ethnic and educational homogamy and a
the homogamy table. After controlling for that,linear change of both forms of homogamy over the

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
INTERMARRIAGE BETWEEN TURKS AND KURDS IN CONTEMPORARY TURKEY 429

Table 5. Log-linear modelselectionfor ethnic and educationalhomogamy (N=11,567 couples)

Model DF BIC

1 ETH*C*S+ETW*C*S+EDUH*C*S+EDUW*C*S 8215.8 46 8028.9


2 1+ETH*ETW 1902.0 45 1719.2
3 1+EDUH*EDUW 6574.2 45 6391.4
41 +ETH*ETW+EDUH*EDUW 109.3 44 -69.5
5 4+ETH*ETW*C 101.2 42 -69.5
6 4+EDUH*EDUW*C 59.4 42 -111.2
74+ETH*ETW*T 101.2 43 -73.5
8 4+EDUH*EDUW*T 60.7 43 -114.0
9 4+ETH*ETW*T+EDUH*EDUW*T 52.4 42 -118.2
10 9+ETH*ETW*S 47.8 41 -118.8
11 9+EDUH*EDUW*S 50.9 41 -115.7
12 9+ETH*ETW*S+EDUH*EDUW*S 46.3 40 -116.3

Note: ETH=Ethnic group husband; ETW=Ethnic group wife; EDUH=Education husband; EDUW=Education wife; C=Cohort; T=Linear
S=Survey.

Table 6. Log-linearparameterswith theirstandarderrorsandantilogsfor (trends in) ethnicandeducationalhomogamy inTurkey on the basisof


Model10 of Table 2

Parameter B SE(B) exp(B) Odds ratio

Ethnic homogamy 1.679** 0.034 5.362 826.63

Trend in ethnic homogamy - 0.102* 0.046 0.903 0.66

Educational homogamy 0.602** 0.017 1.825 11.09

Trend in educational homogamy - 0.150** 0.022 0.861 0.55

*p < 0.05, **p < 0.01.

group. The odds ratio for ethnic homogamy does


cohorts. The better fit of Model 10 compared with
Model 9 indicates that ethnic homogamy differs not differ much from the odds ratio for all couples
in Table 2, which was computed without control for
between the surveys and that this difference should
educational homogamy. This means that ethnic
be taken into account when parameters for ethnic
homogamy in Turkey is only slightly related to
homogamy and its change over time are computed.
educational homogamy.
Besides the models presented in Table 5, we esti-
mated all possible models with higher-orderBoth ethnic homogamy and educational homo-
interactions between ethnicity and educational
gamy decrease significantly over the period
studied. From the first cohort to the second and
level of the partners and the cohort and survey vari-
ables, but none of these models fits the data better the second cohort to the third, the odds ratio
from
than Model 10. of ethnic homogamy decreases each time by about
Table 6 presents the parameters for ethnic and one-third, and the odds ratio of educational homo-
educational homogamy and their trends, computed gamy by 45 per cent. So with respect to both forms of
on the basis of model 10. The odds ratio for ethnic homogamy Turkish society has become more open
homogamy of the multivariate log-linear modelbetween has the 1960s and the 1990s.
a value of 827, which indicates that there is a very
strong tendency towards ethnic homogamy. The
Conclusion
odds ratio for educational homogamy is much
lower. Still, its value of 11.1 indicates a rather strong
A first important conclusion of this study is that
tendency to marry within one's own educational
there were mixed marriages between Kurds and

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
430 GUNDUZ-HOSGOR AND SMITS

Turks inTurkey, but that the tendency towards inter- because the increase in the tendency towards inter-
marriage was rather low. Only 2.4 per cent, or one in marriage was stronger among the Kurdish females
about 40 marriages, was mixed according to the than among the Kurdish males. In the multivariate
ethnic group of the partners. And the odds ratio analysis no gender differences were found.
for ethnic intermarriage had a value of 793, which Turks living in the East of Turkey (which is the
indicates a very strong tendency towards ethnic least developed region and highly occupied by
homogamy. We also found a rather strong tendency Kurds) are much more likely to marry Kurds than
towards educational homogamy in Turkey, but the Turks living in the West and Central/North regions
strength of this tendency is clearly lower than the of Turkey. For the Kurds the percentage of mixed
strength of ethnic homogamy. marriages is lowest in the East and quite high in all
Both ethnic homogamy and educational homo- three other regions. So it seems that the tendency
gamy decrease significantly over the period towards intermarriage is stronger among individuals
studied. From the first cohort to the second cohortwho live in a region where their own group is rela-
and from the second to the third, the odds ratio oftively small and the other group is relatively large.
ethnic homogamy decreases each time by about one- The individual analysis further shows that Kurds
third, and the odds ratio of educational homogamyliving in the urban capitals and largest cities tend to
by 45 per cent. So with respect to both forms ofintermarry significantly more than average, whereas
homogamy, we can conclude that theTurkish society Kurds living in the countryside intermarry signifi-
has become somewhat more open between the 1960scantly less than average. Turks show a similar
and the 1990s. These results are mostly in line withpattern. In urban areas, they intermarry significantly
the 'classical modernization hypothesis', whichmore than in the countryside. So, Turkish-Kurdish
predicts that with the increase of modernization,intermarriage seems to be more a large city pheno-
ethnic and educational homogamy will decrease. menon than a countryside phenomenon. Because
The tendency towards ethnic homogamy remainsindividuals living in large cities on average are
largely intact when it is studied simultaneously withmore 'modern' than individuals living in the coun-
educational homogamy in a multivariate log-linear tryside, this result supports the classical
analysis. This means that the 'by-product hypothesis',modernization hypothesis, which predicts that
which states that the ethnic homogamy of Turks andmodernization will lead to more intermarriage
Kurds in Turkey may be a by-product of educational between Kurds and Turks. However, it is also possi-
homogamy and the large difference in educational ble that the higher intermarriage rates in urban areas
level between Turks and Kurds, cannot be upheld. are due to selective migration of mixed couples to
Besides these rather general results on trends inthe cities.
ethnic and educational homogamy at the aggregate There are large differences in educational level
level, we have also attempted to gain insight into thebetween Turks and Kurds, but the most striking
processes that affect the openness or closure of thedifference is between Kurdish and Turkish women.
group boundaries at the individual level. After all, This difference suggests that Kurds, and especially
the boundaries between groups at the aggregateKurdish women, have less access to development
level are the outcome of the marriage choices of opportunities than Turks. The related fact that
many individual group members. We therefore havemany Kurdish women did not speak Turkish
also studied the effects of individual characteristics (Smits and Gunduz-Ho?g6r, forthcoming) may be
of married respondents and of the place where theya barrier preventing intermarriage between
Kurdish women and Turkish men and, hence, be
lived on the strength of the tendency towards inter-
marriage. responsible for the lower intermarriage levels of
With regard to gender, we found that in the first Kurdish women compared to Kurdish men, in the
marriage cohort Kurdish females tended to be first cohort.
married much less with Turks than Kurdish males. Regarding the relationship between educational
This indicates that, at that time, Kurdish females
level and intermarriage, our findings are very
had little access to the outside world. However,
interesting. Among theTurks, the respondents with
over time this difference became much smaller the lowest educational level tend to marry most with

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
INTERMARRIAGE BETWEEN TURKS AND KURDS IN CONTEMPORARY TURKEY 431

probably begin with increasing the educational


a Kurdish partner, whereas among the Kurds, the
respondents with the highest educational levellevel and widening the horizon of these women.
show the highest intermarriage rates. For both
groups, this effect is quite strong. This finding is in
line with the exchange hypothesis, which states thatNote
'majority men and women may marry a minority
1. Whether education level was entered as a set of
spouse if they gain socio-economically from this'
dummy variables or as an interval variable had hardly
This hypothesis also seems to be supported by the
any effect on the results. For reasons of clarity, the
finding that among the Kurds the upper non-man- interval version was chosen.
uals tend to intermarry most and the farmers the
least. Only the expected gender asymmetry effect in
marital exchange, because of the traditional
References
tendency of females to marry upwards, was not
found in our data. Lack of statistical power, due to
Ahmad, F. (1993) TheMakingofModern Turkey. Routledge,
the presence of very few Kurdish women with more London.

Akcay A.A. (1995) An Empirical Study on Rural


than primary-level education, may be responsible for
this. Structures and Temporary-Seasonal Migration in
Despite the increase in Turkish-Kurdish inter- Southeastern Turkey. PhD thesis (Department of
marriage in recent decades, the number of mixed Sociology, Middle East Technical University,
Ankara, Turkey).
couples is still rather low. In the early 1990s only 3
Aksit; B., Ayata, S., Mutlu, K., Nalbantoglu, H.U.,
per cent of all marriages in which Turkish and/or
Akcay, A.A. and Sen, M. (1996) Population move-
Kurdish partners were involved were mixed accord-
ments in south-eastern Anatolia: some findings of an
ing to ethnic group. Looking only at the smaller empirical research in 1993. New Perspectives on Turkey,
group we see that, at that time, no more than 10 per 14, 53-74.
cent of Kurds marry a partner from the other ethnic Alba, R. and Golden, R.M. (1986) Patterns of ethnic
group. Especially in the East of the country and in marriage in the United States. Social Forces, 65,
the countryside primordial attachments and ethnic 202-223.

consciousness still seem to be very strong. Andrews, P. (1989) Ethnic Groups in the Republic of Turkey.

Besides the low levels of ethnic homogamy Dr. Ludwig Reichert Verlag, Wiesbaden.
among the Kurds, there are also other indications Bell, W. (1981) Modernization. In Encyclopedia of Sociology.
DPG Publishing, Guilford, Conn.
that the traditional tribal structures have not yet dis-
Berkes, N. (1964) The Development of Secularism in Turkey.
appeared completely. There are still substantially
McGill University Press, Montreal.
more network relations (a traditional form of
Blau, P.M. and Duncan, O.D. (1967) The American
organization and solidarity) among the Kurds than OccupationalStructure. Wiley, New York.
among the Turks. For example, in the 1993 TDHS Blau, P.M. and Schwartz, J.E. (1984) Crosscutting Social
data, we found blood relationship within marriage Circles. Academic Press, New York.
to be more common among Kurds (39.2 per cent) Bobrow, B. (1996) Complex insecurity: implications of
than among Turks (20.5 per cent), and arranging a sobering metaphor. International Studies Quarterly,
the marriage themselves instead of via a matchmaker 40/4,435-450.
happens half as much among Kurds (14 per cent) Bruinessen, M.V. (1992) Agha, Shaikh and State. Zed
than among Turks (27.3 per cent). It seems likely Books, London.
Calhoun, C. (1993) Nationalism and ethnicity. Annual
that the large group of illiterate Kurdish women
Review of Sociology, 19, 211-239.
plays a key role in passing the primordial attach-
Cornell, S. and Hartmann, D. (1998) Ethnicity and Race:
ments of the community on to future generations.
Making Identities in a Changing World. Pine Forge
Because they speak little or noTurkish and have little Press, Thousand Oaks, Calif.
access to the outside world, these women depend Davis, K. (1941) Intermarriage in caste societies. A merican
much more on the traditional Kurdish values than
Anthropologist, 43, 388-395.
the more educated women do. Attempts to improveDHS+ (2002). Demographic and Health Surveys (Online).
the situation of the Kurds, therefore, should Available at: http://www.measuredhs.com.

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms
432 GUNDUZ-HO?GOR AND SMITS

Drachsler, J. (1921) Intermarriage in New York City:A Sta- Shavit, Y. and Stier, H. (1994) Ethnicity and education in
tistical Study of the A malgamation of European Peoples. Israel's changing marriage market. Paper presented at
Columbia University Press, New York. the 13th World Congress of the International Socio-
Gordon, M.M. (1964) Assimilation in American Life. logical Association, July 18-23, Bielefeld, Germany.
Oxford University Press, New York. Smits, J. (2000) Ethnisch gemengde huwelijken en Sociale
Giinduz-Hosgor, A. (1997) Development and Women's Cohesie: wat Joegoslavie ons kan leren (Ethnically
Employment Status: Evidence from the Turkish mixed marriage and social cohesion: what we can
Republic 1923-1990. PhD thesis (Department of learn from Yugoslavia). Mens en Maatschappij, 75,
Sociology, University of Western Ontario). 4-21.

Giines-Ayata, A. and Ayata, S. (1998) Ethnicity and Smits, J. and Gunduz-Hosgor, A. (forthcoming) Linguis-
security problems in Turkey. In Lenore, G.M. (ed.) tic capital: language as a socio-economic resource
New Frontiers in the Middle East Security. St. Martin among Kurdish and Arabic women in Turkey. Ethnic
Press, New York. and Racial Studies.

Icduygu, A., David, R. and Sirkeci, i. (1999) The ethnic Smits, J., Ultee, W. and Lammers, J. (1998) Educational
question in an environment of insecurity: the Kurds homogamy in 65 countries: an explanation of differ-
in Turkey. Ethnic and Racial Studies, 22/6, 990-1010. ences in openness using country-level explanatory
Jansen, C. (1982) Inter-ethnic marriages. International variables. A merican Sociological Review, 63, 264-285.
Journal of Comparative Sociology, 23, 225-235. So, A. (1990) Social Change and Development: Modernization,
Kalmijn, M. (1998) Intermarriage and homogamy: Dependency, and World System Theories. Sage, Newbury
causes, patterns, trends. Annual Review of Sociology, Park, Calif.
25, 395-421. SIS (State Institute of Statistics) (1994) Statistical Indi-
Kalmijn, M. (1991) Shifting boundaries: trends in religious cators: 1923-1992. SIS Press, Ankara.
and educational homogamy. American Sociological Tokluoglu, C. (1995) The Formation of the Turkish
Review, 56, 786-800. Nation State and Resistance. PhD thesis (University
Kandiyoti, D. (1988) Bargaining with patriarchy. Gender of Carleton, Ottowa).
andSociety, 2/3, 274-289. Ultee, W., Dessens, J. and Jansen, W. (1988) Why does
Kirisci, K. and Winrow, G. (1997) Kiirt Sorunu, Kokeni ve unemployment come in couples? European Sociological
Gelifimi. (The Kurdish Question: Its Roots and Review, 4, 111-122.
Growth) Tarih Vakf Yurt Yaynlar, Istanbul. Van den Berghe, P.L. (1979) Human Family Systems: An
Lee, S.M. (1988) Intermarriage and ethnic relations in Evolutionary View. Elsevier, New York.
Singapore. Journal of Marriage and the Family, 50, Yegen, M. (1999) Devlet Soyleminde Kiirt Sorunu (The
255-265. Kurdish Question in the State's Discourse). ileti?im
yaynlar, Ankara.
Levi-Strauss, C. (1969) The Elementary Structuresof Kinship.
Beacon Press, Boston.
Merton, R.K. (1941) Intermarriage and social structure:
fact and theory. Psychiatry, 4, 361-374. Author's Address
Mutlu, S. (1996) Ethnic Kurds in Turkey: a demographic
Ay?e Gundiz-Ho?gor, Middle East Technical University,
study. InternationalJournal of Middle East Studies, 28.
Department of Sociology, 06531 Ankara, Turkey.
Raftery, A.E. (1995) Bayesian model selection in social
research. SociologicalMethodology, 25, 111-163. Tel.: 00-90-312-210-5985; fax: 00-90-312-210-1284;
email: hogor@metu.edu.tr
Schoen, R. and Wooldredge, J. (1989) Marriage choices in
North Carolina and Virginia, 1969-71 and 1979-81.
Manuscript received: November 2000.
Journalof Marriage and the Family, 51, 465-481.

This content downloaded from 195.251.133.138 on Tue, 12 Jun 2018 02:56:26 UTC
All use subject to http://about.jstor.org/terms

Das könnte Ihnen auch gefallen